diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS index 41530d4654..7f7cec9cda 100644 --- a/.github/CODEOWNERS +++ b/.github/CODEOWNERS @@ -1,7 +1,7 @@ /web/ui @juliusv /web/ui/module @juliusv @nexucis /storage/remote @cstyan @bwplotka @tomwilkie -/storage/remote/otlptranslator @gouthamve @jesusvazquez +/storage/remote/otlptranslator @aknuds1 @jesusvazquez /discovery/kubernetes @brancz /tsdb @jesusvazquez /promql @roidelapluie diff --git a/.github/dependabot.yml b/.github/dependabot.yml index 9617e04a48..3d56ff2b22 100644 --- a/.github/dependabot.yml +++ b/.github/dependabot.yml @@ -6,11 +6,11 @@ updates: interval: "monthly" groups: k8s.io: - patterns: - - "k8s.io/*" + patterns: + - "k8s.io/*" go.opentelemetry.io: - patterns: - - "go.opentelemetry.io/*" + patterns: + - "go.opentelemetry.io/*" - package-ecosystem: "gomod" directory: "/documentation/examples/remote_storage" schedule: diff --git a/.github/workflows/buf-lint.yml b/.github/workflows/buf-lint.yml index 91de37aa6f..942db6e9b2 100644 --- a/.github/workflows/buf-lint.yml +++ b/.github/workflows/buf-lint.yml @@ -13,7 +13,7 @@ jobs: runs-on: ubuntu-latest steps: - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 - - uses: bufbuild/buf-setup-action@382440cdb8ec7bc25a68d7b4711163d95f7cc3aa # v1.28.1 + - uses: bufbuild/buf-setup-action@517ee23296d5caf38df31c21945e6a54bbc8a89f # v1.30.0 with: github_token: ${{ secrets.GITHUB_TOKEN }} - uses: bufbuild/buf-lint-action@044d13acb1f155179c606aaa2e53aea304d22058 # v1.1.0 diff --git a/.github/workflows/buf.yml b/.github/workflows/buf.yml index 04d4ed8682..9bbfd236e7 100644 --- a/.github/workflows/buf.yml +++ b/.github/workflows/buf.yml @@ -13,7 +13,7 @@ jobs: if: github.repository_owner == 'prometheus' steps: - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 - - uses: bufbuild/buf-setup-action@382440cdb8ec7bc25a68d7b4711163d95f7cc3aa # v1.28.1 + - uses: bufbuild/buf-setup-action@517ee23296d5caf38df31c21945e6a54bbc8a89f # v1.30.0 with: github_token: ${{ secrets.GITHUB_TOKEN }} - uses: bufbuild/buf-lint-action@044d13acb1f155179c606aaa2e53aea304d22058 # v1.1.0 diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 8251f960e9..8866384dba 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -45,7 +45,8 @@ jobs: steps: - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 - run: make build - - run: make test GO_ONLY=1 + # Don't run NPM build; don't run race-detector. + - run: make test GO_ONLY=1 test-flags="" test_ui: name: UI tests @@ -157,11 +158,11 @@ jobs: run: sudo apt-get update && sudo apt-get -y install libsnmp-dev if: github.repository == 'prometheus/snmp_exporter' - name: Lint - uses: golangci/golangci-lint-action@3a919529898de77ec3da873e3063ca4b10e7f5cc # v3.7.0 + uses: golangci/golangci-lint-action@3cfe3a4abbb849e10058ce4af15d205b6da42804 # v4.0.0 with: args: --verbose # Make sure to sync this with Makefile.common and scripts/golangci-lint.yml. - version: v1.55.2 + version: v1.56.2 fuzzing: uses: ./.github/workflows/fuzzing.yml if: github.event_name == 'pull_request' @@ -171,7 +172,7 @@ jobs: publish_main: name: Publish main branch artifacts runs-on: ubuntu-latest - needs: [test_ui, test_go, test_windows, golangci, codeql, build_all] + needs: [test_ui, test_go, test_go_more, test_go_oldest, test_windows, golangci, codeql, build_all] if: github.event_name == 'push' && github.event.ref == 'refs/heads/main' steps: - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 @@ -185,7 +186,7 @@ jobs: publish_release: name: Publish release artefacts runs-on: ubuntu-latest - needs: [test_ui, test_go, test_windows, golangci, codeql, build_all] + needs: [test_ui, test_go, test_go_more, test_go_oldest, test_windows, golangci, codeql, build_all] if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags/v2.') steps: - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 diff --git a/.github/workflows/container_description.yml b/.github/workflows/container_description.yml new file mode 100644 index 0000000000..8a57107dd9 --- /dev/null +++ b/.github/workflows/container_description.yml @@ -0,0 +1,52 @@ +--- +name: Push README to Docker Hub +on: + push: + paths: + - "README.md" + - ".github/workflows/container_description.yml" + branches: [ main, master ] + +permissions: + contents: read + +jobs: + PushDockerHubReadme: + runs-on: ubuntu-latest + name: Push README to Docker Hub + if: github.repository_owner == 'prometheus' || github.repository_owner == 'prometheus-community' # Don't run this workflow on forks. + steps: + - name: git checkout + uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 + - name: Set docker hub repo name + run: echo "DOCKER_REPO_NAME=$(make docker-repo-name)" >> $GITHUB_ENV + - name: Push README to Dockerhub + uses: christian-korneck/update-container-description-action@d36005551adeaba9698d8d67a296bd16fa91f8e8 # v1 + env: + DOCKER_USER: ${{ secrets.DOCKER_HUB_LOGIN }} + DOCKER_PASS: ${{ secrets.DOCKER_HUB_PASSWORD }} + with: + destination_container_repo: ${{ env.DOCKER_REPO_NAME }} + provider: dockerhub + short_description: ${{ env.DOCKER_REPO_NAME }} + readme_file: 'README.md' + + PushQuayIoReadme: + runs-on: ubuntu-latest + name: Push README to quay.io + if: github.repository_owner == 'prometheus' || github.repository_owner == 'prometheus-community' # Don't run this workflow on forks. + steps: + - name: git checkout + uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 + - name: Set quay.io org name + run: echo "DOCKER_REPO=$(echo quay.io/${GITHUB_REPOSITORY_OWNER} | tr -d '-')" >> $GITHUB_ENV + - name: Set quay.io repo name + run: echo "DOCKER_REPO_NAME=$(make docker-repo-name)" >> $GITHUB_ENV + - name: Push README to quay.io + uses: christian-korneck/update-container-description-action@d36005551adeaba9698d8d67a296bd16fa91f8e8 # v1 + env: + DOCKER_APIKEY: ${{ secrets.QUAY_IO_API_TOKEN }} + with: + destination_container_repo: ${{ env.DOCKER_REPO_NAME }} + provider: quay + readme_file: 'README.md' diff --git a/.golangci.yml b/.golangci.yml index 2eeb6c1c8f..f350aed6dd 100644 --- a/.golangci.yml +++ b/.golangci.yml @@ -26,6 +26,7 @@ linters: - testifylint - unconvert - unused + - usestdlibvars issues: max-same-issues: 0 @@ -48,24 +49,24 @@ linters-settings: rules: main: deny: - - pkg: "sync/atomic" - desc: "Use go.uber.org/atomic instead of sync/atomic" - - pkg: "github.com/stretchr/testify/assert" - desc: "Use github.com/stretchr/testify/require instead of github.com/stretchr/testify/assert" - - pkg: "github.com/go-kit/kit/log" - desc: "Use github.com/go-kit/log instead of github.com/go-kit/kit/log" - - pkg: "io/ioutil" - desc: "Use corresponding 'os' or 'io' functions instead." - - pkg: "regexp" - desc: "Use github.com/grafana/regexp instead of regexp" - - pkg: "github.com/pkg/errors" - desc: "Use 'errors' or 'fmt' instead of github.com/pkg/errors" - - pkg: "gzip" - desc: "Use github.com/klauspost/compress instead of gzip" - - pkg: "zlib" - desc: "Use github.com/klauspost/compress instead of zlib" - - pkg: "golang.org/x/exp/slices" - desc: "Use 'slices' instead." + - pkg: "sync/atomic" + desc: "Use go.uber.org/atomic instead of sync/atomic" + - pkg: "github.com/stretchr/testify/assert" + desc: "Use github.com/stretchr/testify/require instead of github.com/stretchr/testify/assert" + - pkg: "github.com/go-kit/kit/log" + desc: "Use github.com/go-kit/log instead of github.com/go-kit/kit/log" + - pkg: "io/ioutil" + desc: "Use corresponding 'os' or 'io' functions instead." + - pkg: "regexp" + desc: "Use github.com/grafana/regexp instead of regexp" + - pkg: "github.com/pkg/errors" + desc: "Use 'errors' or 'fmt' instead of github.com/pkg/errors" + - pkg: "gzip" + desc: "Use github.com/klauspost/compress instead of gzip" + - pkg: "zlib" + desc: "Use github.com/klauspost/compress instead of zlib" + - pkg: "golang.org/x/exp/slices" + desc: "Use 'slices' instead." errcheck: exclude-functions: # Don't flag lines such as "io.Copy(io.Discard, resp.Body)". @@ -135,4 +136,3 @@ linters-settings: - require-error - suite-dont-use-pkg - suite-extra-assert-call - diff --git a/.yamllint b/.yamllint index 955a5a6270..1859cb624b 100644 --- a/.yamllint +++ b/.yamllint @@ -1,5 +1,7 @@ --- extends: default +ignore: | + ui/react-app/node_modules rules: braces: diff --git a/CHANGELOG.md b/CHANGELOG.md index eee024f1df..0afd8d7026 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,5 +1,9 @@ # Changelog +## unreleased + +* [CHANGE] TSDB: Fix the predicate checking for blocks which are beyond the retention period to include the ones right at the retention boundary. #9633 + ## 2.51.2 / 2024-04-09 Bugfix release. diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index 57055ef38c..7687826ba4 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -90,7 +90,7 @@ can modify the `./promql/parser/generated_parser.y.go` manually. ```golang // As of writing this was somewhere around line 600. var ( - yyDebug = 0 // This can be be a number 0 -> 5. + yyDebug = 0 // This can be a number 0 -> 5. yyErrorVerbose = false // This can be set to true. ) diff --git a/MAINTAINERS.md b/MAINTAINERS.md index a776eb3594..630bbdd527 100644 --- a/MAINTAINERS.md +++ b/MAINTAINERS.md @@ -1,7 +1,12 @@ # Maintainers -Julien Pivotto ( / @roidelapluie) and Levi Harrison ( / @LeviHarrison) are the main/default maintainers, some parts of the codebase have other maintainers: +General maintainers: +* Bryan Boreham (bjboreham@gmail.com / @bboreham) +* Levi Harrison (levi@leviharrison.dev / @LeviHarrison) +* Ayoub Mrini (ayoubmrini424@gmail.com / @machine424) +* Julien Pivotto (roidelapluie@prometheus.io / @roidelapluie) +Maintainers for specific parts of the codebase: * `cmd` * `promtool`: David Leadbeater ( / @dgl) * `discovery` @@ -12,6 +17,7 @@ Julien Pivotto ( / @roidelapluie) and Levi Harrison George Krajcsovits ( / @krajorama) * `storage` * `remote`: Callum Styan ( / @cstyan), Bartłomiej Płotka ( / @bwplotka), Tom Wilkie ( / @tomwilkie) + * `otlptranslator`: Arve Knudsen ( / @aknuds1), Jesús Vázquez ( / @jesusvazquez) * `tsdb`: Ganesh Vernekar ( / @codesome), Bartłomiej Płotka ( / @bwplotka), Jesús Vázquez ( / @jesusvazquez) * `agent`: Robert Fratto ( / @rfratto) * `web` @@ -19,7 +25,6 @@ George Krajcsovits ( / @krajorama) * `module`: Augustin Husson ( @nexucis) * `Makefile` and related build configuration: Simon Pasquier ( / @simonpasquier), Ben Kochie ( / @SuperQ) - For the sake of brevity, not all subtrees are explicitly listed. Due to the size of this repository, the natural changes in focus of maintainers over time, and nuances of where particular features live, this list will always be diff --git a/Makefile b/Makefile index ab229f9311..d78813068e 100644 --- a/Makefile +++ b/Makefile @@ -82,7 +82,7 @@ assets-tarball: assets .PHONY: parser parser: @echo ">> running goyacc to generate the .go file." -ifeq (, $(shell command -v goyacc > /dev/null)) +ifeq (, $(shell command -v goyacc 2> /dev/null)) @echo "goyacc not installed so skipping" @echo "To install: go install golang.org/x/tools/cmd/goyacc@v0.6.0" else diff --git a/Makefile.common b/Makefile.common index 92558151e3..0acfb9d806 100644 --- a/Makefile.common +++ b/Makefile.common @@ -49,7 +49,7 @@ endif GOTEST := $(GO) test GOTEST_DIR := ifneq ($(CIRCLE_JOB),) -ifneq ($(shell command -v gotestsum > /dev/null),) +ifneq ($(shell command -v gotestsum 2> /dev/null),) GOTEST_DIR := test-results GOTEST := gotestsum --junitfile $(GOTEST_DIR)/unit-tests.xml -- endif @@ -61,7 +61,7 @@ PROMU_URL := https://github.com/prometheus/promu/releases/download/v$(PROMU_ SKIP_GOLANGCI_LINT := GOLANGCI_LINT := GOLANGCI_LINT_OPTS ?= -GOLANGCI_LINT_VERSION ?= v1.55.2 +GOLANGCI_LINT_VERSION ?= v1.56.2 # golangci-lint only supports linux, darwin and windows platforms on i386/amd64/arm64. # windows isn't included here because of the path separator being different. ifeq ($(GOHOSTOS),$(filter $(GOHOSTOS),linux darwin)) @@ -182,7 +182,7 @@ endif .PHONY: common-yamllint common-yamllint: @echo ">> running yamllint on all YAML files in the repository" -ifeq (, $(shell command -v yamllint > /dev/null)) +ifeq (, $(shell command -v yamllint 2> /dev/null)) @echo "yamllint not installed so skipping" else yamllint . @@ -208,6 +208,10 @@ common-tarball: promu @echo ">> building release tarball" $(PROMU) tarball --prefix $(PREFIX) $(BIN_DIR) +.PHONY: common-docker-repo-name +common-docker-repo-name: + @echo "$(DOCKER_REPO)/$(DOCKER_IMAGE_NAME)" + .PHONY: common-docker $(BUILD_DOCKER_ARCHS) common-docker: $(BUILD_DOCKER_ARCHS) $(BUILD_DOCKER_ARCHS): common-docker-%: diff --git a/RELEASE.md b/RELEASE.md index c2f98ab2ce..f313c4172d 100644 --- a/RELEASE.md +++ b/RELEASE.md @@ -56,6 +56,8 @@ Release cadence of first pre-releases being cut is 6 weeks. | v2.49 | 2023-12-05 | Bartek Plotka (GitHub: @bwplotka) | | v2.50 | 2024-01-16 | Augustin Husson (GitHub: @nexucis) | | v2.51 | 2024-03-07 | Bryan Boreham (GitHub: @bboreham) | +| v2.52 | 2024-04-22 | Arthur Silva Sens (GitHub: @ArthurSens) | +| v2.53 | 2024-06-03 | George Krajcsovits (GitHub: @krajorama) | If you are interested in volunteering please create a pull request against the [prometheus/prometheus](https://github.com/prometheus/prometheus) repository and propose yourself for the release series of your choice. diff --git a/cmd/prometheus/main.go b/cmd/prometheus/main.go index 614d884637..0e15d5ca5f 100644 --- a/cmd/prometheus/main.go +++ b/cmd/prometheus/main.go @@ -447,7 +447,7 @@ func main() { a.Flag("scrape.discovery-reload-interval", "Interval used by scrape manager to throttle target groups updates."). Hidden().Default("5s").SetValue(&cfg.scrape.DiscoveryReloadInterval) - a.Flag("enable-feature", "Comma separated feature names to enable. Valid options: agent, auto-gomemlimit, exemplar-storage, expand-external-labels, memory-snapshot-on-shutdown, promql-at-modifier, promql-negative-offset, promql-per-step-stats, promql-experimental-functions, remote-write-receiver (DEPRECATED), extra-scrape-metrics, new-service-discovery-manager, auto-gomaxprocs, no-default-scrape-port, native-histograms, otlp-write-receiver. See https://prometheus.io/docs/prometheus/latest/feature_flags/ for more details."). + a.Flag("enable-feature", "Comma separated feature names to enable. Valid options: agent, auto-gomemlimit, exemplar-storage, expand-external-labels, memory-snapshot-on-shutdown, promql-per-step-stats, promql-experimental-functions, remote-write-receiver (DEPRECATED), extra-scrape-metrics, new-service-discovery-manager, auto-gomaxprocs, no-default-scrape-port, native-histograms, otlp-write-receiver, created-timestamp-zero-ingestion, concurrent-rule-eval. See https://prometheus.io/docs/prometheus/latest/feature_flags/ for more details."). Default("").StringsVar(&cfg.featureList) promlogflag.AddFlags(a, &cfg.promlogConfig) @@ -960,8 +960,8 @@ func main() { func() error { // Don't forget to release the reloadReady channel so that waiting blocks can exit normally. select { - case <-term: - level.Warn(logger).Log("msg", "Received SIGTERM, exiting gracefully...") + case sig := <-term: + level.Warn(logger).Log("msg", "Received an OS signal, exiting gracefully...", "signal", sig.String()) reloadReady.Close() case <-webHandler.Quit(): level.Warn(logger).Log("msg", "Received termination request via web service, exiting gracefully...") diff --git a/cmd/promtool/main.go b/cmd/promtool/main.go index 47bf02c104..a62ae4fbf4 100644 --- a/cmd/promtool/main.go +++ b/cmd/promtool/main.go @@ -482,7 +482,7 @@ func CheckServerStatus(serverURL *url.URL, checkEndpoint string, roundTripper ht return err } - request, err := http.NewRequest("GET", config.Address, nil) + request, err := http.NewRequest(http.MethodGet, config.Address, nil) if err != nil { return err } diff --git a/cmd/promtool/rules.go b/cmd/promtool/rules.go index d8d6bb83e1..5a18644842 100644 --- a/cmd/promtool/rules.go +++ b/cmd/promtool/rules.go @@ -234,17 +234,3 @@ func (m *multipleAppender) flushAndCommit(ctx context.Context) error { } return nil } - -func max(x, y int64) int64 { - if x > y { - return x - } - return y -} - -func min(x, y int64) int64 { - if x < y { - return x - } - return y -} diff --git a/cmd/promtool/testdata/no-test-group-interval.yml b/cmd/promtool/testdata/no-test-group-interval.yml index d1f6935cd6..99f2ec6467 100644 --- a/cmd/promtool/testdata/no-test-group-interval.yml +++ b/cmd/promtool/testdata/no-test-group-interval.yml @@ -12,4 +12,4 @@ tests: eval_time: 1m exp_samples: - value: 1 - labels: test \ No newline at end of file + labels: test diff --git a/cmd/promtool/tsdb.go b/cmd/promtool/tsdb.go index 73258754e2..b786c92976 100644 --- a/cmd/promtool/tsdb.go +++ b/cmd/promtool/tsdb.go @@ -33,6 +33,7 @@ import ( "github.com/alecthomas/units" "github.com/go-kit/log" + "go.uber.org/atomic" "github.com/prometheus/prometheus/model/labels" "github.com/prometheus/prometheus/promql/parser" @@ -149,8 +150,7 @@ func benchmarkWrite(outPath, samplesFile string, numMetrics, numScrapes int) err } func (b *writeBenchmark) ingestScrapes(lbls []labels.Labels, scrapeCount int) (uint64, error) { - var mu sync.Mutex - var total uint64 + var total atomic.Uint64 for i := 0; i < scrapeCount; i += 100 { var wg sync.WaitGroup @@ -165,22 +165,21 @@ func (b *writeBenchmark) ingestScrapes(lbls []labels.Labels, scrapeCount int) (u wg.Add(1) go func() { + defer wg.Done() + n, err := b.ingestScrapesShard(batch, 100, int64(timeDelta*i)) if err != nil { // exitWithError(err) fmt.Println(" err", err) } - mu.Lock() - total += n - mu.Unlock() - wg.Done() + total.Add(n) }() } wg.Wait() } fmt.Println("ingestion completed") - return total, nil + return total.Load(), nil } func (b *writeBenchmark) ingestScrapesShard(lbls []labels.Labels, scrapeCount int, baset int64) (uint64, error) { diff --git a/cmd/promtool/unittest.go b/cmd/promtool/unittest.go index 4777b88098..6d6683a934 100644 --- a/cmd/promtool/unittest.go +++ b/cmd/promtool/unittest.go @@ -175,13 +175,18 @@ type testGroup struct { } // test performs the unit tests. -func (tg *testGroup) test(evalInterval time.Duration, groupOrderMap map[string]int, queryOpts promql.LazyLoaderOpts, diffFlag bool, ruleFiles ...string) []error { +func (tg *testGroup) test(evalInterval time.Duration, groupOrderMap map[string]int, queryOpts promql.LazyLoaderOpts, diffFlag bool, ruleFiles ...string) (outErr []error) { // Setup testing suite. - suite, err := promql.NewLazyLoader(nil, tg.seriesLoadingString(), queryOpts) + suite, err := promql.NewLazyLoader(tg.seriesLoadingString(), queryOpts) if err != nil { return []error{err} } - defer suite.Close() + defer func() { + err := suite.Close() + if err != nil { + outErr = append(outErr, err) + } + }() suite.SubqueryInterval = evalInterval // Load the rule files. diff --git a/config/config_test.go b/config/config_test.go index 36b125f794..14981d25f0 100644 --- a/config/config_test.go +++ b/config/config_test.go @@ -1840,7 +1840,7 @@ var expectedErrors = []struct { }, { filename: "azure_authentication_method.bad.yml", - errMsg: "unknown authentication_type \"invalid\". Supported types are \"OAuth\" or \"ManagedIdentity\"", + errMsg: "unknown authentication_type \"invalid\". Supported types are \"OAuth\", \"ManagedIdentity\" or \"SDK\"", }, { filename: "azure_bearertoken_basicauth.bad.yml", diff --git a/discovery/azure/azure.go b/discovery/azure/azure.go index 16628c7bfd..7c2ece2c7b 100644 --- a/discovery/azure/azure.go +++ b/discovery/azure/azure.go @@ -65,6 +65,7 @@ const ( azureLabelMachineSize = azureLabel + "machine_size" authMethodOAuth = "OAuth" + authMethodSDK = "SDK" authMethodManagedIdentity = "ManagedIdentity" ) @@ -164,8 +165,8 @@ func (c *SDConfig) UnmarshalYAML(unmarshal func(interface{}) error) error { } } - if c.AuthenticationMethod != authMethodOAuth && c.AuthenticationMethod != authMethodManagedIdentity { - return fmt.Errorf("unknown authentication_type %q. Supported types are %q or %q", c.AuthenticationMethod, authMethodOAuth, authMethodManagedIdentity) + if c.AuthenticationMethod != authMethodOAuth && c.AuthenticationMethod != authMethodManagedIdentity && c.AuthenticationMethod != authMethodSDK { + return fmt.Errorf("unknown authentication_type %q. Supported types are %q, %q or %q", c.AuthenticationMethod, authMethodOAuth, authMethodManagedIdentity, authMethodSDK) } return c.HTTPClientConfig.Validate() @@ -212,6 +213,14 @@ func NewDiscovery(cfg *SDConfig, logger log.Logger, metrics discovery.Discoverer return d, nil } +type client interface { + getVMs(ctx context.Context, resourceGroup string) ([]virtualMachine, error) + getScaleSets(ctx context.Context, resourceGroup string) ([]armcompute.VirtualMachineScaleSet, error) + getScaleSetVMs(ctx context.Context, scaleSet armcompute.VirtualMachineScaleSet) ([]virtualMachine, error) + getVMNetworkInterfaceByID(ctx context.Context, networkInterfaceID string) (*armnetwork.Interface, error) + getVMScaleSetVMNetworkInterfaceByID(ctx context.Context, networkInterfaceID, scaleSetName, instanceID string) (*armnetwork.Interface, error) +} + // azureClient represents multiple Azure Resource Manager providers. type azureClient struct { nic *armnetwork.InterfacesClient @@ -221,14 +230,17 @@ type azureClient struct { logger log.Logger } +var _ client = &azureClient{} + // createAzureClient is a helper function for creating an Azure compute client to ARM. -func createAzureClient(cfg SDConfig) (azureClient, error) { +func createAzureClient(cfg SDConfig, logger log.Logger) (client, error) { cloudConfiguration, err := CloudConfigurationFromName(cfg.Environment) if err != nil { - return azureClient{}, err + return &azureClient{}, err } var c azureClient + c.logger = logger telemetry := policy.TelemetryOptions{ ApplicationID: userAgent, @@ -239,12 +251,12 @@ func createAzureClient(cfg SDConfig) (azureClient, error) { Telemetry: telemetry, }) if err != nil { - return azureClient{}, err + return &azureClient{}, err } client, err := config_util.NewClientFromConfig(cfg.HTTPClientConfig, "azure_sd") if err != nil { - return azureClient{}, err + return &azureClient{}, err } options := &arm.ClientOptions{ ClientOptions: policy.ClientOptions{ @@ -256,25 +268,25 @@ func createAzureClient(cfg SDConfig) (azureClient, error) { c.vm, err = armcompute.NewVirtualMachinesClient(cfg.SubscriptionID, credential, options) if err != nil { - return azureClient{}, err + return &azureClient{}, err } c.nic, err = armnetwork.NewInterfacesClient(cfg.SubscriptionID, credential, options) if err != nil { - return azureClient{}, err + return &azureClient{}, err } c.vmss, err = armcompute.NewVirtualMachineScaleSetsClient(cfg.SubscriptionID, credential, options) if err != nil { - return azureClient{}, err + return &azureClient{}, err } c.vmssvm, err = armcompute.NewVirtualMachineScaleSetVMsClient(cfg.SubscriptionID, credential, options) if err != nil { - return azureClient{}, err + return &azureClient{}, err } - return c, nil + return &c, nil } func newCredential(cfg SDConfig, policyClientOptions policy.ClientOptions) (azcore.TokenCredential, error) { @@ -294,6 +306,16 @@ func newCredential(cfg SDConfig, policyClientOptions policy.ClientOptions) (azco return nil, err } credential = azcore.TokenCredential(secretCredential) + case authMethodSDK: + options := &azidentity.DefaultAzureCredentialOptions{ClientOptions: policyClientOptions} + if len(cfg.TenantID) != 0 { + options.TenantID = cfg.TenantID + } + sdkCredential, err := azidentity.NewDefaultAzureCredential(options) + if err != nil { + return nil, err + } + credential = azcore.TokenCredential(sdkCredential) } return credential, nil } @@ -330,12 +352,11 @@ func newAzureResourceFromID(id string, logger log.Logger) (*arm.ResourceID, erro func (d *Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error) { defer level.Debug(d.logger).Log("msg", "Azure discovery completed") - client, err := createAzureClient(*d.cfg) + client, err := createAzureClient(*d.cfg, d.logger) if err != nil { d.metrics.failuresCount.Inc() return nil, fmt.Errorf("could not create Azure client: %w", err) } - client.logger = d.logger machines, err := client.getVMs(ctx, d.cfg.ResourceGroup) if err != nil { @@ -374,96 +395,8 @@ func (d *Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error) { for _, vm := range machines { go func(vm virtualMachine) { defer wg.Done() - r, err := newAzureResourceFromID(vm.ID, d.logger) - if err != nil { - ch <- target{labelSet: nil, err: err} - return - } - - labels := model.LabelSet{ - azureLabelSubscriptionID: model.LabelValue(d.cfg.SubscriptionID), - azureLabelTenantID: model.LabelValue(d.cfg.TenantID), - azureLabelMachineID: model.LabelValue(vm.ID), - azureLabelMachineName: model.LabelValue(vm.Name), - azureLabelMachineComputerName: model.LabelValue(vm.ComputerName), - azureLabelMachineOSType: model.LabelValue(vm.OsType), - azureLabelMachineLocation: model.LabelValue(vm.Location), - azureLabelMachineResourceGroup: model.LabelValue(r.ResourceGroupName), - azureLabelMachineSize: model.LabelValue(vm.Size), - } - - if vm.ScaleSet != "" { - labels[azureLabelMachineScaleSet] = model.LabelValue(vm.ScaleSet) - } - - for k, v := range vm.Tags { - name := strutil.SanitizeLabelName(k) - labels[azureLabelMachineTag+model.LabelName(name)] = model.LabelValue(*v) - } - - // Get the IP address information via separate call to the network provider. - for _, nicID := range vm.NetworkInterfaces { - var networkInterface *armnetwork.Interface - if v, ok := d.getFromCache(nicID); ok { - networkInterface = v - d.metrics.cacheHitCount.Add(1) - } else { - if vm.ScaleSet == "" { - networkInterface, err = client.getVMNetworkInterfaceByID(ctx, nicID) - } else { - networkInterface, err = client.getVMScaleSetVMNetworkInterfaceByID(ctx, nicID, vm.ScaleSet, vm.InstanceID) - } - - if err != nil { - if errors.Is(err, errorNotFound) { - level.Warn(d.logger).Log("msg", "Network interface does not exist", "name", nicID, "err", err) - } else { - ch <- target{labelSet: nil, err: err} - } - - // Get out of this routine because we cannot continue without a network interface. - return - } - - // Continue processing with the network interface - d.addToCache(nicID, networkInterface) - } - - if networkInterface.Properties == nil { - continue - } - - // Unfortunately Azure does not return information on whether a VM is deallocated. - // This information is available via another API call however the Go SDK does not - // yet support this. On deallocated machines, this value happens to be nil so it - // is a cheap and easy way to determine if a machine is allocated or not. - if networkInterface.Properties.Primary == nil { - level.Debug(d.logger).Log("msg", "Skipping deallocated virtual machine", "machine", vm.Name) - return - } - - if *networkInterface.Properties.Primary { - for _, ip := range networkInterface.Properties.IPConfigurations { - // IPAddress is a field defined in PublicIPAddressPropertiesFormat, - // therefore we need to validate that both are not nil. - if ip.Properties != nil && ip.Properties.PublicIPAddress != nil && ip.Properties.PublicIPAddress.Properties != nil && ip.Properties.PublicIPAddress.Properties.IPAddress != nil { - labels[azureLabelMachinePublicIP] = model.LabelValue(*ip.Properties.PublicIPAddress.Properties.IPAddress) - } - if ip.Properties != nil && ip.Properties.PrivateIPAddress != nil { - labels[azureLabelMachinePrivateIP] = model.LabelValue(*ip.Properties.PrivateIPAddress) - address := net.JoinHostPort(*ip.Properties.PrivateIPAddress, fmt.Sprintf("%d", d.port)) - labels[model.AddressLabel] = model.LabelValue(address) - ch <- target{labelSet: labels, err: nil} - return - } - // If we made it here, we don't have a private IP which should be impossible. - // Return an empty target and error to ensure an all or nothing situation. - err = fmt.Errorf("unable to find a private IP for VM %s", vm.Name) - ch <- target{labelSet: nil, err: err} - return - } - } - } + labelSet, err := d.vmToLabelSet(ctx, client, vm) + ch <- target{labelSet: labelSet, err: err} }(vm) } @@ -484,6 +417,95 @@ func (d *Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error) { return []*targetgroup.Group{&tg}, nil } +func (d *Discovery) vmToLabelSet(ctx context.Context, client client, vm virtualMachine) (model.LabelSet, error) { + r, err := newAzureResourceFromID(vm.ID, d.logger) + if err != nil { + return nil, err + } + + labels := model.LabelSet{ + azureLabelSubscriptionID: model.LabelValue(d.cfg.SubscriptionID), + azureLabelTenantID: model.LabelValue(d.cfg.TenantID), + azureLabelMachineID: model.LabelValue(vm.ID), + azureLabelMachineName: model.LabelValue(vm.Name), + azureLabelMachineComputerName: model.LabelValue(vm.ComputerName), + azureLabelMachineOSType: model.LabelValue(vm.OsType), + azureLabelMachineLocation: model.LabelValue(vm.Location), + azureLabelMachineResourceGroup: model.LabelValue(r.ResourceGroupName), + azureLabelMachineSize: model.LabelValue(vm.Size), + } + + if vm.ScaleSet != "" { + labels[azureLabelMachineScaleSet] = model.LabelValue(vm.ScaleSet) + } + + for k, v := range vm.Tags { + name := strutil.SanitizeLabelName(k) + labels[azureLabelMachineTag+model.LabelName(name)] = model.LabelValue(*v) + } + + // Get the IP address information via separate call to the network provider. + for _, nicID := range vm.NetworkInterfaces { + var networkInterface *armnetwork.Interface + if v, ok := d.getFromCache(nicID); ok { + networkInterface = v + d.metrics.cacheHitCount.Add(1) + } else { + if vm.ScaleSet == "" { + networkInterface, err = client.getVMNetworkInterfaceByID(ctx, nicID) + } else { + networkInterface, err = client.getVMScaleSetVMNetworkInterfaceByID(ctx, nicID, vm.ScaleSet, vm.InstanceID) + } + if err != nil { + if errors.Is(err, errorNotFound) { + level.Warn(d.logger).Log("msg", "Network interface does not exist", "name", nicID, "err", err) + } else { + return nil, err + } + // Get out of this routine because we cannot continue without a network interface. + return nil, nil + } + + // Continue processing with the network interface + d.addToCache(nicID, networkInterface) + } + + if networkInterface.Properties == nil { + continue + } + + // Unfortunately Azure does not return information on whether a VM is deallocated. + // This information is available via another API call however the Go SDK does not + // yet support this. On deallocated machines, this value happens to be nil so it + // is a cheap and easy way to determine if a machine is allocated or not. + if networkInterface.Properties.Primary == nil { + level.Debug(d.logger).Log("msg", "Skipping deallocated virtual machine", "machine", vm.Name) + return nil, nil + } + + if *networkInterface.Properties.Primary { + for _, ip := range networkInterface.Properties.IPConfigurations { + // IPAddress is a field defined in PublicIPAddressPropertiesFormat, + // therefore we need to validate that both are not nil. + if ip.Properties != nil && ip.Properties.PublicIPAddress != nil && ip.Properties.PublicIPAddress.Properties != nil && ip.Properties.PublicIPAddress.Properties.IPAddress != nil { + labels[azureLabelMachinePublicIP] = model.LabelValue(*ip.Properties.PublicIPAddress.Properties.IPAddress) + } + if ip.Properties != nil && ip.Properties.PrivateIPAddress != nil { + labels[azureLabelMachinePrivateIP] = model.LabelValue(*ip.Properties.PrivateIPAddress) + address := net.JoinHostPort(*ip.Properties.PrivateIPAddress, fmt.Sprintf("%d", d.port)) + labels[model.AddressLabel] = model.LabelValue(address) + return labels, nil + } + // If we made it here, we don't have a private IP which should be impossible. + // Return an empty target and error to ensure an all or nothing situation. + return nil, fmt.Errorf("unable to find a private IP for VM %s", vm.Name) + } + } + } + // TODO: Should we say something at this point? + return nil, nil +} + func (client *azureClient) getVMs(ctx context.Context, resourceGroup string) ([]virtualMachine, error) { var vms []virtualMachine if len(resourceGroup) == 0 { diff --git a/discovery/azure/azure_test.go b/discovery/azure/azure_test.go index 1e437c75f2..32dab66c8c 100644 --- a/discovery/azure/azure_test.go +++ b/discovery/azure/azure_test.go @@ -14,16 +14,24 @@ package azure import ( + "context" + "fmt" "testing" "github.com/Azure/azure-sdk-for-go/sdk/azcore/arm" "github.com/Azure/azure-sdk-for-go/sdk/resourcemanager/compute/armcompute/v5" + "github.com/Azure/azure-sdk-for-go/sdk/resourcemanager/network/armnetwork/v4" + cache "github.com/Code-Hex/go-generics-cache" + "github.com/Code-Hex/go-generics-cache/policy/lru" + "github.com/go-kit/log" "github.com/stretchr/testify/require" "go.uber.org/goleak" ) func TestMain(m *testing.M) { - goleak.VerifyTestMain(m) + goleak.VerifyTestMain(m, + goleak.IgnoreTopFunction("github.com/Code-Hex/go-generics-cache.(*janitor).run.func1"), + ) } func TestMapFromVMWithEmptyTags(t *testing.T) { @@ -79,6 +87,91 @@ func TestMapFromVMWithEmptyTags(t *testing.T) { require.Equal(t, expectedVM, actualVM) } +func TestVMToLabelSet(t *testing.T) { + id := "/subscriptions/00000000-0000-0000-0000-000000000000/test" + name := "name" + size := "size" + vmSize := armcompute.VirtualMachineSizeTypes(size) + osType := armcompute.OperatingSystemTypesLinux + vmType := "type" + location := "westeurope" + computerName := "computer_name" + networkID := "/subscriptions/00000000-0000-0000-0000-000000000000/network1" + ipAddress := "10.20.30.40" + primary := true + networkProfile := armcompute.NetworkProfile{ + NetworkInterfaces: []*armcompute.NetworkInterfaceReference{ + { + ID: &networkID, + Properties: &armcompute.NetworkInterfaceReferenceProperties{Primary: &primary}, + }, + }, + } + properties := &armcompute.VirtualMachineProperties{ + OSProfile: &armcompute.OSProfile{ + ComputerName: &computerName, + }, + StorageProfile: &armcompute.StorageProfile{ + OSDisk: &armcompute.OSDisk{ + OSType: &osType, + }, + }, + NetworkProfile: &networkProfile, + HardwareProfile: &armcompute.HardwareProfile{ + VMSize: &vmSize, + }, + } + + testVM := armcompute.VirtualMachine{ + ID: &id, + Name: &name, + Type: &vmType, + Location: &location, + Tags: nil, + Properties: properties, + } + + expectedVM := virtualMachine{ + ID: id, + Name: name, + ComputerName: computerName, + Type: vmType, + Location: location, + OsType: "Linux", + Tags: map[string]*string{}, + NetworkInterfaces: []string{networkID}, + Size: size, + } + + actualVM := mapFromVM(testVM) + + require.Equal(t, expectedVM, actualVM) + + cfg := DefaultSDConfig + d := &Discovery{ + cfg: &cfg, + logger: log.NewNopLogger(), + cache: cache.New(cache.AsLRU[string, *armnetwork.Interface](lru.WithCapacity(5))), + } + network := armnetwork.Interface{ + Name: &networkID, + Properties: &armnetwork.InterfacePropertiesFormat{ + Primary: &primary, + IPConfigurations: []*armnetwork.InterfaceIPConfiguration{ + {Properties: &armnetwork.InterfaceIPConfigurationPropertiesFormat{ + PrivateIPAddress: &ipAddress, + }}, + }, + }, + } + client := &mockAzureClient{ + networkInterface: &network, + } + labelSet, err := d.vmToLabelSet(context.Background(), client, actualVM) + require.NoError(t, err) + require.Len(t, labelSet, 11) +} + func TestMapFromVMWithEmptyOSType(t *testing.T) { id := "test" name := "name" @@ -381,3 +474,35 @@ func TestNewAzureResourceFromID(t *testing.T) { require.Equal(t, tc.expected.ResourceGroupName, actual.ResourceGroupName) } } + +type mockAzureClient struct { + networkInterface *armnetwork.Interface +} + +var _ client = &mockAzureClient{} + +func (*mockAzureClient) getVMs(ctx context.Context, resourceGroup string) ([]virtualMachine, error) { + return nil, nil +} + +func (*mockAzureClient) getScaleSets(ctx context.Context, resourceGroup string) ([]armcompute.VirtualMachineScaleSet, error) { + return nil, nil +} + +func (*mockAzureClient) getScaleSetVMs(ctx context.Context, scaleSet armcompute.VirtualMachineScaleSet) ([]virtualMachine, error) { + return nil, nil +} + +func (m *mockAzureClient) getVMNetworkInterfaceByID(ctx context.Context, networkInterfaceID string) (*armnetwork.Interface, error) { + if networkInterfaceID == "" { + return nil, fmt.Errorf("parameter networkInterfaceID cannot be empty") + } + return m.networkInterface, nil +} + +func (m *mockAzureClient) getVMScaleSetVMNetworkInterfaceByID(ctx context.Context, networkInterfaceID, scaleSetName, instanceID string) (*armnetwork.Interface, error) { + if scaleSetName == "" { + return nil, fmt.Errorf("parameter virtualMachineScaleSetName cannot be empty") + } + return m.networkInterface, nil +} diff --git a/discovery/eureka/client.go b/discovery/eureka/client.go index a833415a53..52e8ce7b48 100644 --- a/discovery/eureka/client.go +++ b/discovery/eureka/client.go @@ -81,7 +81,7 @@ const appListPath string = "/apps" func fetchApps(ctx context.Context, server string, client *http.Client) (*Applications, error) { url := fmt.Sprintf("%s%s", server, appListPath) - request, err := http.NewRequest("GET", url, nil) + request, err := http.NewRequest(http.MethodGet, url, nil) if err != nil { return nil, err } diff --git a/discovery/file/file_test.go b/discovery/file/file_test.go index 521a3c0f16..179ac5cd1c 100644 --- a/discovery/file/file_test.go +++ b/discovery/file/file_test.go @@ -208,7 +208,6 @@ func (t *testRunner) requireUpdate(ref time.Time, expected []*targetgroup.Group) select { case <-timeout: t.Fatalf("Expected update but got none") - return case <-time.After(defaultWait / 10): if ref.Equal(t.lastReceive()) { // No update received. diff --git a/discovery/hetzner/robot.go b/discovery/hetzner/robot.go index 1d8aa9302f..2cf6006f80 100644 --- a/discovery/hetzner/robot.go +++ b/discovery/hetzner/robot.go @@ -70,7 +70,7 @@ func newRobotDiscovery(conf *SDConfig, _ log.Logger) (*robotDiscovery, error) { } func (d *robotDiscovery) refresh(context.Context) ([]*targetgroup.Group, error) { - req, err := http.NewRequest("GET", d.endpoint+"/server", nil) + req, err := http.NewRequest(http.MethodGet, d.endpoint+"/server", nil) if err != nil { return nil, err } diff --git a/discovery/http/http.go b/discovery/http/http.go index 8dd21ec9e4..ff76fd7627 100644 --- a/discovery/http/http.go +++ b/discovery/http/http.go @@ -150,7 +150,7 @@ func NewDiscovery(conf *SDConfig, logger log.Logger, clientOpts []config.HTTPCli } func (d *Discovery) Refresh(ctx context.Context) ([]*targetgroup.Group, error) { - req, err := http.NewRequest("GET", d.url, nil) + req, err := http.NewRequest(http.MethodGet, d.url, nil) if err != nil { return nil, err } diff --git a/discovery/kubernetes/kubernetes.go b/discovery/kubernetes/kubernetes.go index d6b8115848..a8b6f85899 100644 --- a/discovery/kubernetes/kubernetes.go +++ b/discovery/kubernetes/kubernetes.go @@ -311,7 +311,7 @@ func New(l log.Logger, metrics discovery.DiscovererMetrics, conf *SDConfig) (*Di } case conf.APIServer.URL == nil: // Use the Kubernetes provided pod service account - // as described in https://kubernetes.io/docs/admin/service-accounts-admin/ + // as described in https://kubernetes.io/docs/tasks/run-application/access-api-from-pod/#using-official-client-libraries kcfg, err = rest.InClusterConfig() if err != nil { return nil, err @@ -485,8 +485,8 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) { eps := NewEndpointSlice( log.With(d.logger, "role", "endpointslice"), informer, - cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), - cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled), + d.mustNewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), + d.mustNewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled), nodeInf, d.metrics.eventCount, ) @@ -545,8 +545,8 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) { eps := NewEndpoints( log.With(d.logger, "role", "endpoint"), d.newEndpointsByNodeInformer(elw), - cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), - cache.NewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled), + d.mustNewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), + d.mustNewSharedInformer(plw, &apiv1.Pod{}, resyncDisabled), nodeInf, d.metrics.eventCount, ) @@ -602,7 +602,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) { } svc := NewService( log.With(d.logger, "role", "service"), - cache.NewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), + d.mustNewSharedInformer(slw, &apiv1.Service{}, resyncDisabled), d.metrics.eventCount, ) d.discoverers = append(d.discoverers, svc) @@ -641,7 +641,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) { return i.Watch(ctx, options) }, } - informer = cache.NewSharedInformer(ilw, &networkv1.Ingress{}, resyncDisabled) + informer = d.mustNewSharedInformer(ilw, &networkv1.Ingress{}, resyncDisabled) } else { i := d.client.NetworkingV1beta1().Ingresses(namespace) ilw := &cache.ListWatch{ @@ -656,7 +656,7 @@ func (d *Discovery) Run(ctx context.Context, ch chan<- []*targetgroup.Group) { return i.Watch(ctx, options) }, } - informer = cache.NewSharedInformer(ilw, &v1beta1.Ingress{}, resyncDisabled) + informer = d.mustNewSharedInformer(ilw, &v1beta1.Ingress{}, resyncDisabled) } ingress := NewIngress( log.With(d.logger, "role", "ingress"), @@ -747,7 +747,7 @@ func (d *Discovery) newNodeInformer(ctx context.Context) cache.SharedInformer { return d.client.CoreV1().Nodes().Watch(ctx, options) }, } - return cache.NewSharedInformer(nlw, &apiv1.Node{}, resyncDisabled) + return d.mustNewSharedInformer(nlw, &apiv1.Node{}, resyncDisabled) } func (d *Discovery) newPodsByNodeInformer(plw *cache.ListWatch) cache.SharedIndexInformer { @@ -762,7 +762,7 @@ func (d *Discovery) newPodsByNodeInformer(plw *cache.ListWatch) cache.SharedInde } } - return cache.NewSharedIndexInformer(plw, &apiv1.Pod{}, resyncDisabled, indexers) + return d.mustNewSharedIndexInformer(plw, &apiv1.Pod{}, resyncDisabled, indexers) } func (d *Discovery) newEndpointsByNodeInformer(plw *cache.ListWatch) cache.SharedIndexInformer { @@ -783,7 +783,7 @@ func (d *Discovery) newEndpointsByNodeInformer(plw *cache.ListWatch) cache.Share return pods, nil } if !d.attachMetadata.Node { - return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers) + return d.mustNewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers) } indexers[nodeIndex] = func(obj interface{}) ([]string, error) { @@ -809,13 +809,13 @@ func (d *Discovery) newEndpointsByNodeInformer(plw *cache.ListWatch) cache.Share return nodes, nil } - return cache.NewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers) + return d.mustNewSharedIndexInformer(plw, &apiv1.Endpoints{}, resyncDisabled, indexers) } func (d *Discovery) newEndpointSlicesByNodeInformer(plw *cache.ListWatch, object runtime.Object) cache.SharedIndexInformer { indexers := make(map[string]cache.IndexFunc) if !d.attachMetadata.Node { - return cache.NewSharedIndexInformer(plw, object, resyncDisabled, indexers) + return d.mustNewSharedIndexInformer(plw, object, resyncDisabled, indexers) } indexers[nodeIndex] = func(obj interface{}) ([]string, error) { @@ -854,7 +854,32 @@ func (d *Discovery) newEndpointSlicesByNodeInformer(plw *cache.ListWatch, object return nodes, nil } - return cache.NewSharedIndexInformer(plw, object, resyncDisabled, indexers) + return d.mustNewSharedIndexInformer(plw, object, resyncDisabled, indexers) +} + +func (d *Discovery) informerWatchErrorHandler(r *cache.Reflector, err error) { + d.metrics.failuresCount.Inc() + cache.DefaultWatchErrorHandler(r, err) +} + +func (d *Discovery) mustNewSharedInformer(lw cache.ListerWatcher, exampleObject runtime.Object, defaultEventHandlerResyncPeriod time.Duration) cache.SharedInformer { + informer := cache.NewSharedInformer(lw, exampleObject, defaultEventHandlerResyncPeriod) + // Invoking SetWatchErrorHandler should fail only if the informer has been started beforehand. + // Such a scenario would suggest an incorrect use of the API, thus the panic. + if err := informer.SetWatchErrorHandler(d.informerWatchErrorHandler); err != nil { + panic(err) + } + return informer +} + +func (d *Discovery) mustNewSharedIndexInformer(lw cache.ListerWatcher, exampleObject runtime.Object, defaultEventHandlerResyncPeriod time.Duration, indexers cache.Indexers) cache.SharedIndexInformer { + informer := cache.NewSharedIndexInformer(lw, exampleObject, defaultEventHandlerResyncPeriod, indexers) + // Invoking SetWatchErrorHandler should fail only if the informer has been started beforehand. + // Such a scenario would suggest an incorrect use of the API, thus the panic. + if err := informer.SetWatchErrorHandler(d.informerWatchErrorHandler); err != nil { + panic(err) + } + return informer } func checkDiscoveryV1Supported(client kubernetes.Interface) (bool, error) { diff --git a/discovery/kubernetes/kubernetes_test.go b/discovery/kubernetes/kubernetes_test.go index e3dd093009..552f8a4453 100644 --- a/discovery/kubernetes/kubernetes_test.go +++ b/discovery/kubernetes/kubernetes_test.go @@ -21,12 +21,16 @@ import ( "time" "github.com/go-kit/log" + prom_testutil "github.com/prometheus/client_golang/prometheus/testutil" "github.com/stretchr/testify/require" + apierrors "k8s.io/apimachinery/pkg/api/errors" "k8s.io/apimachinery/pkg/runtime" "k8s.io/apimachinery/pkg/version" + "k8s.io/apimachinery/pkg/watch" fakediscovery "k8s.io/client-go/discovery/fake" "k8s.io/client-go/kubernetes" "k8s.io/client-go/kubernetes/fake" + kubetesting "k8s.io/client-go/testing" "k8s.io/client-go/tools/cache" "github.com/prometheus/client_golang/prometheus" @@ -314,3 +318,39 @@ func TestCheckNetworkingV1Supported(t *testing.T) { }) } } + +func TestFailuresCountMetric(t *testing.T) { + tests := []struct { + role Role + minFailedWatches int + }{ + {RoleNode, 1}, + {RolePod, 1}, + {RoleService, 1}, + {RoleEndpoint, 3}, + {RoleEndpointSlice, 3}, + {RoleIngress, 1}, + } + + for _, tc := range tests { + tc := tc + t.Run(string(tc.role), func(t *testing.T) { + t.Parallel() + + n, c := makeDiscovery(tc.role, NamespaceDiscovery{}) + // The counter is initialized and no failures at the beginning. + require.Equal(t, float64(0), prom_testutil.ToFloat64(n.metrics.failuresCount)) + + // Simulate an error on watch requests. + c.Discovery().(*fakediscovery.FakeDiscovery).PrependWatchReactor("*", func(action kubetesting.Action) (bool, watch.Interface, error) { + return true, nil, apierrors.NewUnauthorized("unauthorized") + }) + + // Start the discovery. + k8sDiscoveryTest{discovery: n}.Run(t) + + // At least the errors of the initial watches should be caught (watches are retried on errors). + require.GreaterOrEqual(t, prom_testutil.ToFloat64(n.metrics.failuresCount), float64(tc.minFailedWatches)) + }) + } +} diff --git a/discovery/kubernetes/metrics.go b/discovery/kubernetes/metrics.go index 7d384fb96a..fe419bc782 100644 --- a/discovery/kubernetes/metrics.go +++ b/discovery/kubernetes/metrics.go @@ -22,7 +22,8 @@ import ( var _ discovery.DiscovererMetrics = (*kubernetesMetrics)(nil) type kubernetesMetrics struct { - eventCount *prometheus.CounterVec + eventCount *prometheus.CounterVec + failuresCount prometheus.Counter metricRegisterer discovery.MetricRegisterer } @@ -37,10 +38,18 @@ func newDiscovererMetrics(reg prometheus.Registerer, rmi discovery.RefreshMetric }, []string{"role", "event"}, ), + failuresCount: prometheus.NewCounter( + prometheus.CounterOpts{ + Namespace: discovery.KubernetesMetricsNamespace, + Name: "failures_total", + Help: "The number of failed WATCH/LIST requests.", + }, + ), } m.metricRegisterer = discovery.NewMetricRegisterer(reg, []prometheus.Collector{ m.eventCount, + m.failuresCount, }) // Initialize metric vectors. @@ -61,6 +70,8 @@ func newDiscovererMetrics(reg prometheus.Registerer, rmi discovery.RefreshMetric } } + m.failuresCount.Add(0) + return m } diff --git a/discovery/legacymanager/manager_test.go b/discovery/legacymanager/manager_test.go index 6fbecabc2a..1ed699645d 100644 --- a/discovery/legacymanager/manager_test.go +++ b/discovery/legacymanager/manager_test.go @@ -733,7 +733,6 @@ func verifyPresence(t *testing.T, tSets map[poolKey]map[string]*targetgroup.Grou t.Helper() if _, ok := tSets[poolKey]; !ok { t.Fatalf("'%s' should be present in Pool keys: %v", poolKey, tSets) - return } match := false diff --git a/discovery/linode/linode.go b/discovery/linode/linode.go index 94f0a63bbb..2a5475b854 100644 --- a/discovery/linode/linode.go +++ b/discovery/linode/linode.go @@ -59,17 +59,22 @@ const ( linodeLabelSpecsVCPUs = linodeLabel + "specs_vcpus" linodeLabelSpecsTransferBytes = linodeLabel + "specs_transfer_bytes" linodeLabelExtraIPs = linodeLabel + "extra_ips" + linodeLabelIPv6Ranges = linodeLabel + "ipv6_ranges" // This is our events filter; when polling for changes, we care only about // events since our last refresh. - // Docs: https://www.linode.com/docs/api/account/#events-list + // Docs: https://www.linode.com/docs/api/account/#events-list. filterTemplate = `{"created": {"+gte": "%s"}}` + + // Optional region filtering. + regionFilterTemplate = `{"region": "%s"}` ) // DefaultSDConfig is the default Linode SD configuration. var DefaultSDConfig = SDConfig{ TagSeparator: ",", Port: 80, + Region: "", RefreshInterval: model.Duration(60 * time.Second), HTTPClientConfig: config.DefaultHTTPClientConfig, } @@ -85,6 +90,7 @@ type SDConfig struct { RefreshInterval model.Duration `yaml:"refresh_interval"` Port int `yaml:"port"` TagSeparator string `yaml:"tag_separator,omitempty"` + Region string `yaml:"region,omitempty"` } // NewDiscovererMetrics implements discovery.Config. @@ -122,6 +128,7 @@ type Discovery struct { *refresh.Discovery client *linodego.Client port int + region string tagSeparator string lastRefreshTimestamp time.Time pollCount int @@ -139,6 +146,7 @@ func NewDiscovery(conf *SDConfig, logger log.Logger, metrics discovery.Discovere d := &Discovery{ port: conf.Port, + region: conf.Region, tagSeparator: conf.TagSeparator, pollCount: 0, lastRefreshTimestamp: time.Now().UTC(), @@ -224,16 +232,31 @@ func (d *Discovery) refreshData(ctx context.Context) ([]*targetgroup.Group, erro tg := &targetgroup.Group{ Source: "Linode", } + opts := linodego.ListOptions{ + PageSize: 500, + } + + // If region filter provided, use it to constrain results. + if d.region != "" { + opts.Filter = fmt.Sprintf(regionFilterTemplate, d.region) + } // Gather all linode instances. - instances, err := d.client.ListInstances(ctx, &linodego.ListOptions{PageSize: 500}) + instances, err := d.client.ListInstances(ctx, &opts) if err != nil { d.metrics.failuresCount.Inc() return nil, err } // Gather detailed IP address info for all IPs on all linode instances. - detailedIPs, err := d.client.ListIPAddresses(ctx, &linodego.ListOptions{PageSize: 500}) + detailedIPs, err := d.client.ListIPAddresses(ctx, &opts) + if err != nil { + d.metrics.failuresCount.Inc() + return nil, err + } + + // Gather detailed IPv6 Range info for all linode instances. + ipv6RangeList, err := d.client.ListIPv6Ranges(ctx, &opts) if err != nil { d.metrics.failuresCount.Inc() return nil, err @@ -248,7 +271,7 @@ func (d *Discovery) refreshData(ctx context.Context) ([]*targetgroup.Group, erro privateIPv4, publicIPv4, publicIPv6 string privateIPv4RDNS, publicIPv4RDNS, publicIPv6RDNS string backupsStatus string - extraIPs []string + extraIPs, ipv6Ranges []string ) for _, ip := range instance.IPv4 { @@ -276,17 +299,23 @@ func (d *Discovery) refreshData(ctx context.Context) ([]*targetgroup.Group, erro } if instance.IPv6 != "" { + slaac := strings.Split(instance.IPv6, "/")[0] for _, detailedIP := range detailedIPs { - if detailedIP.Address != strings.Split(instance.IPv6, "/")[0] { + if detailedIP.Address != slaac { continue } - publicIPv6 = detailedIP.Address if detailedIP.RDNS != "" && detailedIP.RDNS != "null" { publicIPv6RDNS = detailedIP.RDNS } } + for _, ipv6Range := range ipv6RangeList { + if ipv6Range.RouteTarget != slaac { + continue + } + ipv6Ranges = append(ipv6Ranges, fmt.Sprintf("%s/%d", ipv6Range.Range, ipv6Range.Prefix)) + } } if instance.Backups.Enabled { @@ -330,12 +359,20 @@ func (d *Discovery) refreshData(ctx context.Context) ([]*targetgroup.Group, erro if len(extraIPs) > 0 { // This instance has more than one of at least one type of IP address (public, private, - // IPv4, IPv6, etc. We provide those extra IPs found here just like we do for instance + // IPv4,etc. We provide those extra IPs found here just like we do for instance // tags, we surround a separated list with the tagSeparator config. ips := d.tagSeparator + strings.Join(extraIPs, d.tagSeparator) + d.tagSeparator labels[linodeLabelExtraIPs] = model.LabelValue(ips) } + if len(ipv6Ranges) > 0 { + // This instance has more than one IPv6 Ranges routed to it we provide these + // Ranges found here just like we do for instance tags, we surround a separated + // list with the tagSeparator config. + ips := d.tagSeparator + strings.Join(ipv6Ranges, d.tagSeparator) + d.tagSeparator + labels[linodeLabelIPv6Ranges] = model.LabelValue(ips) + } + tg.Targets = append(tg.Targets, labels) } return []*targetgroup.Group{tg}, nil diff --git a/discovery/linode/linode_test.go b/discovery/linode/linode_test.go index a6a16d82aa..3c10650653 100644 --- a/discovery/linode/linode_test.go +++ b/discovery/linode/linode_test.go @@ -28,159 +28,236 @@ import ( "github.com/prometheus/prometheus/discovery" ) -type LinodeSDTestSuite struct { - Mock *SDMock -} - -func (s *LinodeSDTestSuite) TearDownSuite() { - s.Mock.ShutdownServer() -} - -func (s *LinodeSDTestSuite) SetupTest(t *testing.T) { - s.Mock = NewSDMock(t) - s.Mock.Setup() - - s.Mock.HandleLinodeInstancesList() - s.Mock.HandleLinodeNeworkingIPs() - s.Mock.HandleLinodeAccountEvents() -} - func TestLinodeSDRefresh(t *testing.T) { - sdmock := &LinodeSDTestSuite{} - sdmock.SetupTest(t) - t.Cleanup(sdmock.TearDownSuite) + sdmock := NewSDMock(t) + sdmock.Setup() - cfg := DefaultSDConfig - cfg.HTTPClientConfig.Authorization = &config.Authorization{ - Credentials: tokenID, - Type: "Bearer", + tests := map[string]struct { + region string + targetCount int + want []model.LabelSet + }{ + "no_region": {region: "", targetCount: 4, want: []model.LabelSet{ + { + "__address__": model.LabelValue("45.33.82.151:80"), + "__meta_linode_instance_id": model.LabelValue("26838044"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-1"), + "__meta_linode_image": model.LabelValue("linode/arch"), + "__meta_linode_private_ipv4": model.LabelValue("192.168.170.51"), + "__meta_linode_public_ipv4": model.LabelValue("45.33.82.151"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:1382"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li1028-151.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("us-east"), + "__meta_linode_type": model.LabelValue("g6-standard-2"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("85899345920"), + "__meta_linode_specs_memory_bytes": model.LabelValue("4294967296"), + "__meta_linode_specs_vcpus": model.LabelValue("2"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("4194304000"), + "__meta_linode_extra_ips": model.LabelValue(",96.126.108.16,192.168.201.25,"), + }, + { + "__address__": model.LabelValue("139.162.196.43:80"), + "__meta_linode_instance_id": model.LabelValue("26848419"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-2"), + "__meta_linode_image": model.LabelValue("linode/debian10"), + "__meta_linode_private_ipv4": model.LabelValue(""), + "__meta_linode_public_ipv4": model.LabelValue("139.162.196.43"), + "__meta_linode_public_ipv6": model.LabelValue("2a01:7e00::f03c:92ff:fe1a:9976"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li1359-43.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("eu-west"), + "__meta_linode_type": model.LabelValue("g6-standard-2"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("85899345920"), + "__meta_linode_specs_memory_bytes": model.LabelValue("4294967296"), + "__meta_linode_specs_vcpus": model.LabelValue("2"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("4194304000"), + }, + { + "__address__": model.LabelValue("192.53.120.25:80"), + "__meta_linode_instance_id": model.LabelValue("26837938"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-3"), + "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), + "__meta_linode_private_ipv4": model.LabelValue(""), + "__meta_linode_public_ipv4": model.LabelValue("192.53.120.25"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c04::f03c:92ff:fe1a:fb68"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li2216-25.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("ca-central"), + "__meta_linode_type": model.LabelValue("g6-standard-1"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("53687091200"), + "__meta_linode_specs_memory_bytes": model.LabelValue("2147483648"), + "__meta_linode_specs_vcpus": model.LabelValue("1"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("2097152000"), + "__meta_linode_ipv6_ranges": model.LabelValue(",2600:3c04:e001:456::/64,"), + }, + { + "__address__": model.LabelValue("66.228.47.103:80"), + "__meta_linode_instance_id": model.LabelValue("26837992"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-4"), + "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), + "__meta_linode_private_ipv4": model.LabelValue("192.168.148.94"), + "__meta_linode_public_ipv4": model.LabelValue("66.228.47.103"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:fb4c"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li328-103.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("us-east"), + "__meta_linode_type": model.LabelValue("g6-nanode-1"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("26843545600"), + "__meta_linode_specs_memory_bytes": model.LabelValue("1073741824"), + "__meta_linode_specs_vcpus": model.LabelValue("1"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("1048576000"), + "__meta_linode_extra_ips": model.LabelValue(",172.104.18.104,"), + "__meta_linode_ipv6_ranges": model.LabelValue(",2600:3c03:e000:123::/64,"), + }, + }}, + "us-east": {region: "us-east", targetCount: 2, want: []model.LabelSet{ + { + "__address__": model.LabelValue("45.33.82.151:80"), + "__meta_linode_instance_id": model.LabelValue("26838044"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-1"), + "__meta_linode_image": model.LabelValue("linode/arch"), + "__meta_linode_private_ipv4": model.LabelValue("192.168.170.51"), + "__meta_linode_public_ipv4": model.LabelValue("45.33.82.151"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:1382"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li1028-151.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("us-east"), + "__meta_linode_type": model.LabelValue("g6-standard-2"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("85899345920"), + "__meta_linode_specs_memory_bytes": model.LabelValue("4294967296"), + "__meta_linode_specs_vcpus": model.LabelValue("2"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("4194304000"), + "__meta_linode_extra_ips": model.LabelValue(",96.126.108.16,192.168.201.25,"), + }, + { + "__address__": model.LabelValue("66.228.47.103:80"), + "__meta_linode_instance_id": model.LabelValue("26837992"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-4"), + "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), + "__meta_linode_private_ipv4": model.LabelValue("192.168.148.94"), + "__meta_linode_public_ipv4": model.LabelValue("66.228.47.103"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:fb4c"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li328-103.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("us-east"), + "__meta_linode_type": model.LabelValue("g6-nanode-1"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("26843545600"), + "__meta_linode_specs_memory_bytes": model.LabelValue("1073741824"), + "__meta_linode_specs_vcpus": model.LabelValue("1"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("1048576000"), + "__meta_linode_extra_ips": model.LabelValue(",172.104.18.104,"), + "__meta_linode_ipv6_ranges": model.LabelValue(",2600:3c03:e000:123::/64,"), + }, + }}, + "us-central": {region: "ca-central", targetCount: 1, want: []model.LabelSet{ + { + "__address__": model.LabelValue("192.53.120.25:80"), + "__meta_linode_instance_id": model.LabelValue("26837938"), + "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-3"), + "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), + "__meta_linode_private_ipv4": model.LabelValue(""), + "__meta_linode_public_ipv4": model.LabelValue("192.53.120.25"), + "__meta_linode_public_ipv6": model.LabelValue("2600:3c04::f03c:92ff:fe1a:fb68"), + "__meta_linode_private_ipv4_rdns": model.LabelValue(""), + "__meta_linode_public_ipv4_rdns": model.LabelValue("li2216-25.members.linode.com"), + "__meta_linode_public_ipv6_rdns": model.LabelValue(""), + "__meta_linode_region": model.LabelValue("ca-central"), + "__meta_linode_type": model.LabelValue("g6-standard-1"), + "__meta_linode_status": model.LabelValue("running"), + "__meta_linode_tags": model.LabelValue(",monitoring,"), + "__meta_linode_group": model.LabelValue(""), + "__meta_linode_gpus": model.LabelValue("0"), + "__meta_linode_hypervisor": model.LabelValue("kvm"), + "__meta_linode_backups": model.LabelValue("disabled"), + "__meta_linode_specs_disk_bytes": model.LabelValue("53687091200"), + "__meta_linode_specs_memory_bytes": model.LabelValue("2147483648"), + "__meta_linode_specs_vcpus": model.LabelValue("1"), + "__meta_linode_specs_transfer_bytes": model.LabelValue("2097152000"), + "__meta_linode_ipv6_ranges": model.LabelValue(",2600:3c04:e001:456::/64,"), + }, + }}, } - reg := prometheus.NewRegistry() - refreshMetrics := discovery.NewRefreshMetrics(reg) - metrics := cfg.NewDiscovererMetrics(reg, refreshMetrics) - require.NoError(t, metrics.Register()) - defer metrics.Unregister() - defer refreshMetrics.Unregister() + for _, tc := range tests { + cfg := DefaultSDConfig + if tc.region != "" { + cfg.Region = tc.region + } + cfg.HTTPClientConfig.Authorization = &config.Authorization{ + Credentials: tokenID, + Type: "Bearer", + } - d, err := NewDiscovery(&cfg, log.NewNopLogger(), metrics) - require.NoError(t, err) - endpoint, err := url.Parse(sdmock.Mock.Endpoint()) - require.NoError(t, err) - d.client.SetBaseURL(endpoint.String()) + reg := prometheus.NewRegistry() + refreshMetrics := discovery.NewRefreshMetrics(reg) + metrics := cfg.NewDiscovererMetrics(reg, refreshMetrics) + require.NoError(t, metrics.Register()) + defer metrics.Unregister() + defer refreshMetrics.Unregister() - tgs, err := d.refresh(context.Background()) - require.NoError(t, err) + d, err := NewDiscovery(&cfg, log.NewNopLogger(), metrics) + require.NoError(t, err) + endpoint, err := url.Parse(sdmock.Endpoint()) + require.NoError(t, err) + d.client.SetBaseURL(endpoint.String()) - require.Len(t, tgs, 1) + tgs, err := d.refresh(context.Background()) + require.NoError(t, err) - tg := tgs[0] - require.NotNil(t, tg) - require.NotNil(t, tg.Targets) - require.Len(t, tg.Targets, 4) + require.Len(t, tgs, 1) - for i, lbls := range []model.LabelSet{ - { - "__address__": model.LabelValue("45.33.82.151:80"), - "__meta_linode_instance_id": model.LabelValue("26838044"), - "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-1"), - "__meta_linode_image": model.LabelValue("linode/arch"), - "__meta_linode_private_ipv4": model.LabelValue("192.168.170.51"), - "__meta_linode_public_ipv4": model.LabelValue("45.33.82.151"), - "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:1382"), - "__meta_linode_private_ipv4_rdns": model.LabelValue(""), - "__meta_linode_public_ipv4_rdns": model.LabelValue("li1028-151.members.linode.com"), - "__meta_linode_public_ipv6_rdns": model.LabelValue(""), - "__meta_linode_region": model.LabelValue("us-east"), - "__meta_linode_type": model.LabelValue("g6-standard-2"), - "__meta_linode_status": model.LabelValue("running"), - "__meta_linode_tags": model.LabelValue(",monitoring,"), - "__meta_linode_group": model.LabelValue(""), - "__meta_linode_gpus": model.LabelValue("0"), - "__meta_linode_hypervisor": model.LabelValue("kvm"), - "__meta_linode_backups": model.LabelValue("disabled"), - "__meta_linode_specs_disk_bytes": model.LabelValue("85899345920"), - "__meta_linode_specs_memory_bytes": model.LabelValue("4294967296"), - "__meta_linode_specs_vcpus": model.LabelValue("2"), - "__meta_linode_specs_transfer_bytes": model.LabelValue("4194304000"), - "__meta_linode_extra_ips": model.LabelValue(",96.126.108.16,192.168.201.25,"), - }, - { - "__address__": model.LabelValue("139.162.196.43:80"), - "__meta_linode_instance_id": model.LabelValue("26848419"), - "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-2"), - "__meta_linode_image": model.LabelValue("linode/debian10"), - "__meta_linode_private_ipv4": model.LabelValue(""), - "__meta_linode_public_ipv4": model.LabelValue("139.162.196.43"), - "__meta_linode_public_ipv6": model.LabelValue("2a01:7e00::f03c:92ff:fe1a:9976"), - "__meta_linode_private_ipv4_rdns": model.LabelValue(""), - "__meta_linode_public_ipv4_rdns": model.LabelValue("li1359-43.members.linode.com"), - "__meta_linode_public_ipv6_rdns": model.LabelValue(""), - "__meta_linode_region": model.LabelValue("eu-west"), - "__meta_linode_type": model.LabelValue("g6-standard-2"), - "__meta_linode_status": model.LabelValue("running"), - "__meta_linode_tags": model.LabelValue(",monitoring,"), - "__meta_linode_group": model.LabelValue(""), - "__meta_linode_gpus": model.LabelValue("0"), - "__meta_linode_hypervisor": model.LabelValue("kvm"), - "__meta_linode_backups": model.LabelValue("disabled"), - "__meta_linode_specs_disk_bytes": model.LabelValue("85899345920"), - "__meta_linode_specs_memory_bytes": model.LabelValue("4294967296"), - "__meta_linode_specs_vcpus": model.LabelValue("2"), - "__meta_linode_specs_transfer_bytes": model.LabelValue("4194304000"), - }, - { - "__address__": model.LabelValue("192.53.120.25:80"), - "__meta_linode_instance_id": model.LabelValue("26837938"), - "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-3"), - "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), - "__meta_linode_private_ipv4": model.LabelValue(""), - "__meta_linode_public_ipv4": model.LabelValue("192.53.120.25"), - "__meta_linode_public_ipv6": model.LabelValue("2600:3c04::f03c:92ff:fe1a:fb68"), - "__meta_linode_private_ipv4_rdns": model.LabelValue(""), - "__meta_linode_public_ipv4_rdns": model.LabelValue("li2216-25.members.linode.com"), - "__meta_linode_public_ipv6_rdns": model.LabelValue(""), - "__meta_linode_region": model.LabelValue("ca-central"), - "__meta_linode_type": model.LabelValue("g6-standard-1"), - "__meta_linode_status": model.LabelValue("running"), - "__meta_linode_tags": model.LabelValue(",monitoring,"), - "__meta_linode_group": model.LabelValue(""), - "__meta_linode_gpus": model.LabelValue("0"), - "__meta_linode_hypervisor": model.LabelValue("kvm"), - "__meta_linode_backups": model.LabelValue("disabled"), - "__meta_linode_specs_disk_bytes": model.LabelValue("53687091200"), - "__meta_linode_specs_memory_bytes": model.LabelValue("2147483648"), - "__meta_linode_specs_vcpus": model.LabelValue("1"), - "__meta_linode_specs_transfer_bytes": model.LabelValue("2097152000"), - }, - { - "__address__": model.LabelValue("66.228.47.103:80"), - "__meta_linode_instance_id": model.LabelValue("26837992"), - "__meta_linode_instance_label": model.LabelValue("prometheus-linode-sd-exporter-4"), - "__meta_linode_image": model.LabelValue("linode/ubuntu20.04"), - "__meta_linode_private_ipv4": model.LabelValue("192.168.148.94"), - "__meta_linode_public_ipv4": model.LabelValue("66.228.47.103"), - "__meta_linode_public_ipv6": model.LabelValue("2600:3c03::f03c:92ff:fe1a:fb4c"), - "__meta_linode_private_ipv4_rdns": model.LabelValue(""), - "__meta_linode_public_ipv4_rdns": model.LabelValue("li328-103.members.linode.com"), - "__meta_linode_public_ipv6_rdns": model.LabelValue(""), - "__meta_linode_region": model.LabelValue("us-east"), - "__meta_linode_type": model.LabelValue("g6-nanode-1"), - "__meta_linode_status": model.LabelValue("running"), - "__meta_linode_tags": model.LabelValue(",monitoring,"), - "__meta_linode_group": model.LabelValue(""), - "__meta_linode_gpus": model.LabelValue("0"), - "__meta_linode_hypervisor": model.LabelValue("kvm"), - "__meta_linode_backups": model.LabelValue("disabled"), - "__meta_linode_specs_disk_bytes": model.LabelValue("26843545600"), - "__meta_linode_specs_memory_bytes": model.LabelValue("1073741824"), - "__meta_linode_specs_vcpus": model.LabelValue("1"), - "__meta_linode_specs_transfer_bytes": model.LabelValue("1048576000"), - "__meta_linode_extra_ips": model.LabelValue(",172.104.18.104,"), - }, - } { - t.Run(fmt.Sprintf("item %d", i), func(t *testing.T) { - require.Equal(t, lbls, tg.Targets[i]) - }) + tg := tgs[0] + require.NotNil(t, tg) + require.NotNil(t, tg.Targets) + require.Len(t, tg.Targets, tc.targetCount) + + for i, lbls := range tc.want { + t.Run(fmt.Sprintf("item %d", i), func(t *testing.T) { + require.Equal(t, lbls, tg.Targets[i]) + }) + } } } diff --git a/discovery/linode/mock_test.go b/discovery/linode/mock_test.go index ea0e8e0a8b..50f0572ecd 100644 --- a/discovery/linode/mock_test.go +++ b/discovery/linode/mock_test.go @@ -14,12 +14,17 @@ package linode import ( + "encoding/json" "fmt" "net/http" "net/http/httptest" + "os" + "path/filepath" "testing" ) +const tokenID = "7b2c56dd51edd90952c1b94c472b94b176f20c5c777e376849edd8ad1c6c03bb" + // SDMock is the interface for the Linode mock. type SDMock struct { t *testing.T @@ -43,412 +48,34 @@ func (m *SDMock) Endpoint() string { func (m *SDMock) Setup() { m.Mux = http.NewServeMux() m.Server = httptest.NewServer(m.Mux) + m.t.Cleanup(m.Server.Close) + m.SetupHandlers() } -// ShutdownServer creates the mock server. -func (m *SDMock) ShutdownServer() { - m.Server.Close() -} - -const tokenID = "7b2c56dd51edd90952c1b94c472b94b176f20c5c777e376849edd8ad1c6c03bb" - -// HandleLinodeInstancesList mocks linode instances list. -func (m *SDMock) HandleLinodeInstancesList() { - m.Mux.HandleFunc("/v4/linode/instances", func(w http.ResponseWriter, r *http.Request) { - if r.Header.Get("Authorization") != fmt.Sprintf("Bearer %s", tokenID) { - w.WriteHeader(http.StatusUnauthorized) - return - } - - w.Header().Set("content-type", "application/json; charset=utf-8") - w.WriteHeader(http.StatusOK) - - fmt.Fprint(w, ` -{ - "data": [ - { - "id": 26838044, - "label": "prometheus-linode-sd-exporter-1", - "group": "", - "status": "running", - "created": "2021-05-12T04:23:44", - "updated": "2021-05-12T04:23:44", - "type": "g6-standard-2", - "ipv4": [ - "45.33.82.151", - "96.126.108.16", - "192.168.170.51", - "192.168.201.25" - ], - "ipv6": "2600:3c03::f03c:92ff:fe1a:1382/128", - "image": "linode/arch", - "region": "us-east", - "specs": { - "disk": 81920, - "memory": 4096, - "vcpus": 2, - "gpus": 0, - "transfer": 4000 - }, - "alerts": { - "cpu": 180, - "network_in": 10, - "network_out": 10, - "transfer_quota": 80, - "io": 10000 - }, - "backups": { - "enabled": false, - "schedule": { - "day": null, - "window": null - }, - "last_successful": null - }, - "hypervisor": "kvm", - "watchdog_enabled": true, - "tags": [ - "monitoring" - ] - }, - { - "id": 26848419, - "label": "prometheus-linode-sd-exporter-2", - "group": "", - "status": "running", - "created": "2021-05-12T12:41:49", - "updated": "2021-05-12T12:41:49", - "type": "g6-standard-2", - "ipv4": [ - "139.162.196.43" - ], - "ipv6": "2a01:7e00::f03c:92ff:fe1a:9976/128", - "image": "linode/debian10", - "region": "eu-west", - "specs": { - "disk": 81920, - "memory": 4096, - "vcpus": 2, - "gpus": 0, - "transfer": 4000 - }, - "alerts": { - "cpu": 180, - "network_in": 10, - "network_out": 10, - "transfer_quota": 80, - "io": 10000 - }, - "backups": { - "enabled": false, - "schedule": { - "day": null, - "window": null - }, - "last_successful": null - }, - "hypervisor": "kvm", - "watchdog_enabled": true, - "tags": [ - "monitoring" - ] - }, - { - "id": 26837938, - "label": "prometheus-linode-sd-exporter-3", - "group": "", - "status": "running", - "created": "2021-05-12T04:20:11", - "updated": "2021-05-12T04:20:11", - "type": "g6-standard-1", - "ipv4": [ - "192.53.120.25" - ], - "ipv6": "2600:3c04::f03c:92ff:fe1a:fb68/128", - "image": "linode/ubuntu20.04", - "region": "ca-central", - "specs": { - "disk": 51200, - "memory": 2048, - "vcpus": 1, - "gpus": 0, - "transfer": 2000 - }, - "alerts": { - "cpu": 90, - "network_in": 10, - "network_out": 10, - "transfer_quota": 80, - "io": 10000 - }, - "backups": { - "enabled": false, - "schedule": { - "day": null, - "window": null - }, - "last_successful": null - }, - "hypervisor": "kvm", - "watchdog_enabled": true, - "tags": [ - "monitoring" - ] - }, - { - "id": 26837992, - "label": "prometheus-linode-sd-exporter-4", - "group": "", - "status": "running", - "created": "2021-05-12T04:22:06", - "updated": "2021-05-12T04:22:06", - "type": "g6-nanode-1", - "ipv4": [ - "66.228.47.103", - "172.104.18.104", - "192.168.148.94" - ], - "ipv6": "2600:3c03::f03c:92ff:fe1a:fb4c/128", - "image": "linode/ubuntu20.04", - "region": "us-east", - "specs": { - "disk": 25600, - "memory": 1024, - "vcpus": 1, - "gpus": 0, - "transfer": 1000 - }, - "alerts": { - "cpu": 90, - "network_in": 10, - "network_out": 10, - "transfer_quota": 80, - "io": 10000 - }, - "backups": { - "enabled": false, - "schedule": { - "day": null, - "window": null - }, - "last_successful": null - }, - "hypervisor": "kvm", - "watchdog_enabled": true, - "tags": [ - "monitoring" - ] - } - ], - "page": 1, - "pages": 1, - "results": 4 -}`, - ) - }) -} - -// HandleLinodeNeworkingIPs mocks linode networking ips endpoint. -func (m *SDMock) HandleLinodeNeworkingIPs() { - m.Mux.HandleFunc("/v4/networking/ips", func(w http.ResponseWriter, r *http.Request) { - if r.Header.Get("Authorization") != fmt.Sprintf("Bearer %s", tokenID) { - w.WriteHeader(http.StatusUnauthorized) - return - } - - w.Header().Set("content-type", "application/json; charset=utf-8") - w.WriteHeader(http.StatusOK) - - fmt.Fprint(w, ` -{ - "page": 1, - "pages": 1, - "results": 13, - "data": [ - { - "address": "192.53.120.25", - "gateway": "192.53.120.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li2216-25.members.linode.com", - "linode_id": 26837938, - "region": "ca-central" - }, - { - "address": "66.228.47.103", - "gateway": "66.228.47.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li328-103.members.linode.com", - "linode_id": 26837992, - "region": "us-east" - }, - { - "address": "172.104.18.104", - "gateway": "172.104.18.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li1832-104.members.linode.com", - "linode_id": 26837992, - "region": "us-east" - }, - { - "address": "192.168.148.94", - "gateway": null, - "subnet_mask": "255.255.128.0", - "prefix": 17, - "type": "ipv4", - "public": false, - "rdns": null, - "linode_id": 26837992, - "region": "us-east" - }, - { - "address": "192.168.170.51", - "gateway": null, - "subnet_mask": "255.255.128.0", - "prefix": 17, - "type": "ipv4", - "public": false, - "rdns": null, - "linode_id": 26838044, - "region": "us-east" - }, - { - "address": "96.126.108.16", - "gateway": "96.126.108.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li365-16.members.linode.com", - "linode_id": 26838044, - "region": "us-east" - }, - { - "address": "45.33.82.151", - "gateway": "45.33.82.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li1028-151.members.linode.com", - "linode_id": 26838044, - "region": "us-east" - }, - { - "address": "192.168.201.25", - "gateway": null, - "subnet_mask": "255.255.128.0", - "prefix": 17, - "type": "ipv4", - "public": false, - "rdns": null, - "linode_id": 26838044, - "region": "us-east" - }, - { - "address": "139.162.196.43", - "gateway": "139.162.196.1", - "subnet_mask": "255.255.255.0", - "prefix": 24, - "type": "ipv4", - "public": true, - "rdns": "li1359-43.members.linode.com", - "linode_id": 26848419, - "region": "eu-west" - }, - { - "address": "2600:3c04::f03c:92ff:fe1a:fb68", - "gateway": "fe80::1", - "subnet_mask": "ffff:ffff:ffff:ffff::", - "prefix": 64, - "type": "ipv6", - "rdns": null, - "linode_id": 26837938, - "region": "ca-central", - "public": true - }, - { - "address": "2600:3c03::f03c:92ff:fe1a:fb4c", - "gateway": "fe80::1", - "subnet_mask": "ffff:ffff:ffff:ffff::", - "prefix": 64, - "type": "ipv6", - "rdns": null, - "linode_id": 26837992, - "region": "us-east", - "public": true - }, - { - "address": "2600:3c03::f03c:92ff:fe1a:1382", - "gateway": "fe80::1", - "subnet_mask": "ffff:ffff:ffff:ffff::", - "prefix": 64, - "type": "ipv6", - "rdns": null, - "linode_id": 26838044, - "region": "us-east", - "public": true - }, - { - "address": "2a01:7e00::f03c:92ff:fe1a:9976", - "gateway": "fe80::1", - "subnet_mask": "ffff:ffff:ffff:ffff::", - "prefix": 64, - "type": "ipv6", - "rdns": null, - "linode_id": 26848419, - "region": "eu-west", - "public": true - } - ] -}`, - ) - }) -} - -// HandleLinodeAccountEvents mocks linode the account/events endpoint. -func (m *SDMock) HandleLinodeAccountEvents() { - m.Mux.HandleFunc("/v4/account/events", func(w http.ResponseWriter, r *http.Request) { - if r.Header.Get("Authorization") != fmt.Sprintf("Bearer %s", tokenID) { - w.WriteHeader(http.StatusUnauthorized) - return - } - - if r.Header.Get("X-Filter") == "" { - // This should never happen; if the client sends an events request without - // a filter, cause it to fail. The error below is not a real response from - // the API, but should aid in debugging failed tests. - w.WriteHeader(http.StatusBadRequest) - fmt.Fprint(w, ` -{ - "errors": [ - { - "reason": "Request missing expected X-Filter headers" - } - ] -}`, - ) - return - } - - w.Header().Set("content-type", "application/json; charset=utf-8") - w.WriteHeader(http.StatusOK) - - fmt.Fprint(w, ` -{ - "data": [], - "results": 0, - "pages": 1, - "page": 1 -}`, - ) - }) +// SetupHandlers for endpoints of interest. +func (m *SDMock) SetupHandlers() { + for _, handler := range []string{"/v4/account/events", "/v4/linode/instances", "/v4/networking/ips", "/v4/networking/ipv6/ranges"} { + m.Mux.HandleFunc(handler, func(w http.ResponseWriter, r *http.Request) { + if r.Header.Get("Authorization") != fmt.Sprintf("Bearer %s", tokenID) { + w.WriteHeader(http.StatusUnauthorized) + return + } + xFilter := struct { + Region string `json:"region"` + }{} + json.Unmarshal([]byte(r.Header.Get("X-Filter")), &xFilter) + + directory := "testdata/no_region_filter" + if xFilter.Region != "" { // Validate region filter matches test criteria. + directory = "testdata/" + xFilter.Region + } + if response, err := os.ReadFile(filepath.Join(directory, r.URL.Path+".json")); err == nil { + w.Header().Add("content-type", "application/json; charset=utf-8") + w.WriteHeader(http.StatusOK) + w.Write(response) + return + } + w.WriteHeader(http.StatusInternalServerError) + }) + } } diff --git a/discovery/linode/testdata/ca-central/v4/account/events.json b/discovery/linode/testdata/ca-central/v4/account/events.json new file mode 100644 index 0000000000..ca302e4fd0 --- /dev/null +++ b/discovery/linode/testdata/ca-central/v4/account/events.json @@ -0,0 +1,6 @@ +{ + "data": [], + "results": 0, + "pages": 1, + "page": 1 +} diff --git a/discovery/linode/testdata/ca-central/v4/linode/instances.json b/discovery/linode/testdata/ca-central/v4/linode/instances.json new file mode 100644 index 0000000000..dfc1172477 --- /dev/null +++ b/discovery/linode/testdata/ca-central/v4/linode/instances.json @@ -0,0 +1,49 @@ +{ + "data": [ + { + "id": 26837938, + "label": "prometheus-linode-sd-exporter-3", + "group": "", + "status": "running", + "created": "2021-05-12T04:20:11", + "updated": "2021-05-12T04:20:11", + "type": "g6-standard-1", + "ipv4": [ + "192.53.120.25" + ], + "ipv6": "2600:3c04::f03c:92ff:fe1a:fb68/128", + "image": "linode/ubuntu20.04", + "region": "ca-central", + "specs": { + "disk": 51200, + "memory": 2048, + "vcpus": 1, + "gpus": 0, + "transfer": 2000 + }, + "alerts": { + "cpu": 90, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + } + ], + "page": 1, + "pages": 1, + "results": 1 +} diff --git a/discovery/linode/testdata/ca-central/v4/networking/ips.json b/discovery/linode/testdata/ca-central/v4/networking/ips.json new file mode 100644 index 0000000000..23d974a886 --- /dev/null +++ b/discovery/linode/testdata/ca-central/v4/networking/ips.json @@ -0,0 +1,29 @@ +{ + "page": 1, + "pages": 1, + "results": 2, + "data": [ + { + "address": "192.53.120.25", + "gateway": "192.53.120.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li2216-25.members.linode.com", + "linode_id": 26837938, + "region": "ca-central" + }, + { + "address": "2600:3c04::f03c:92ff:fe1a:fb68", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26837938, + "region": "ca-central", + "public": true + } + ] +} diff --git a/discovery/linode/testdata/ca-central/v4/networking/ipv6/ranges.json b/discovery/linode/testdata/ca-central/v4/networking/ipv6/ranges.json new file mode 100644 index 0000000000..442615cbbc --- /dev/null +++ b/discovery/linode/testdata/ca-central/v4/networking/ipv6/ranges.json @@ -0,0 +1,13 @@ +{ + "data": [ + { + "range": "2600:3c04:e001:456::", + "prefix": 64, + "region": "ca-central", + "route_target": "2600:3c04::f03c:92ff:fe1a:fb68" + } + ], + "page": 1, + "pages": 1, + "results": 1 +} diff --git a/discovery/linode/testdata/no_region_filter/v4/account/events.json b/discovery/linode/testdata/no_region_filter/v4/account/events.json new file mode 100644 index 0000000000..ca302e4fd0 --- /dev/null +++ b/discovery/linode/testdata/no_region_filter/v4/account/events.json @@ -0,0 +1,6 @@ +{ + "data": [], + "results": 0, + "pages": 1, + "page": 1 +} diff --git a/discovery/linode/testdata/no_region_filter/v4/linode/instances.json b/discovery/linode/testdata/no_region_filter/v4/linode/instances.json new file mode 100644 index 0000000000..25d5271d9c --- /dev/null +++ b/discovery/linode/testdata/no_region_filter/v4/linode/instances.json @@ -0,0 +1,180 @@ +{ + "data": [ + { + "id": 26838044, + "label": "prometheus-linode-sd-exporter-1", + "group": "", + "status": "running", + "created": "2021-05-12T04:23:44", + "updated": "2021-05-12T04:23:44", + "type": "g6-standard-2", + "ipv4": [ + "45.33.82.151", + "96.126.108.16", + "192.168.170.51", + "192.168.201.25" + ], + "ipv6": "2600:3c03::f03c:92ff:fe1a:1382/128", + "image": "linode/arch", + "region": "us-east", + "specs": { + "disk": 81920, + "memory": 4096, + "vcpus": 2, + "gpus": 0, + "transfer": 4000 + }, + "alerts": { + "cpu": 180, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + }, + { + "id": 26848419, + "label": "prometheus-linode-sd-exporter-2", + "group": "", + "status": "running", + "created": "2021-05-12T12:41:49", + "updated": "2021-05-12T12:41:49", + "type": "g6-standard-2", + "ipv4": [ + "139.162.196.43" + ], + "ipv6": "2a01:7e00::f03c:92ff:fe1a:9976/128", + "image": "linode/debian10", + "region": "eu-west", + "specs": { + "disk": 81920, + "memory": 4096, + "vcpus": 2, + "gpus": 0, + "transfer": 4000 + }, + "alerts": { + "cpu": 180, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + }, + { + "id": 26837938, + "label": "prometheus-linode-sd-exporter-3", + "group": "", + "status": "running", + "created": "2021-05-12T04:20:11", + "updated": "2021-05-12T04:20:11", + "type": "g6-standard-1", + "ipv4": [ + "192.53.120.25" + ], + "ipv6": "2600:3c04::f03c:92ff:fe1a:fb68/128", + "image": "linode/ubuntu20.04", + "region": "ca-central", + "specs": { + "disk": 51200, + "memory": 2048, + "vcpus": 1, + "gpus": 0, + "transfer": 2000 + }, + "alerts": { + "cpu": 90, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + }, + { + "id": 26837992, + "label": "prometheus-linode-sd-exporter-4", + "group": "", + "status": "running", + "created": "2021-05-12T04:22:06", + "updated": "2021-05-12T04:22:06", + "type": "g6-nanode-1", + "ipv4": [ + "66.228.47.103", + "172.104.18.104", + "192.168.148.94" + ], + "ipv6": "2600:3c03::f03c:92ff:fe1a:fb4c/128", + "image": "linode/ubuntu20.04", + "region": "us-east", + "specs": { + "disk": 25600, + "memory": 1024, + "vcpus": 1, + "gpus": 0, + "transfer": 1000 + }, + "alerts": { + "cpu": 90, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + } + ], + "page": 1, + "pages": 1, + "results": 4 +} diff --git a/discovery/linode/testdata/no_region_filter/v4/networking/ips.json b/discovery/linode/testdata/no_region_filter/v4/networking/ips.json new file mode 100644 index 0000000000..5173036f1c --- /dev/null +++ b/discovery/linode/testdata/no_region_filter/v4/networking/ips.json @@ -0,0 +1,150 @@ +{ + "page": 1, + "pages": 1, + "results": 13, + "data": [ + { + "address": "192.53.120.25", + "gateway": "192.53.120.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li2216-25.members.linode.com", + "linode_id": 26837938, + "region": "ca-central" + }, + { + "address": "66.228.47.103", + "gateway": "66.228.47.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li328-103.members.linode.com", + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "172.104.18.104", + "gateway": "172.104.18.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li1832-104.members.linode.com", + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "192.168.148.94", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "192.168.170.51", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "96.126.108.16", + "gateway": "96.126.108.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li365-16.members.linode.com", + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "45.33.82.151", + "gateway": "45.33.82.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li1028-151.members.linode.com", + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "192.168.201.25", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "139.162.196.43", + "gateway": "139.162.196.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li1359-43.members.linode.com", + "linode_id": 26848419, + "region": "eu-west" + }, + { + "address": "2600:3c04::f03c:92ff:fe1a:fb68", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26837938, + "region": "ca-central", + "public": true + }, + { + "address": "2600:3c03::f03c:92ff:fe1a:fb4c", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26837992, + "region": "us-east", + "public": true + }, + { + "address": "2600:3c03::f03c:92ff:fe1a:1382", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26838044, + "region": "us-east", + "public": true + }, + { + "address": "2a01:7e00::f03c:92ff:fe1a:9976", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26848419, + "region": "eu-west", + "public": true + } + ] +} diff --git a/discovery/linode/testdata/no_region_filter/v4/networking/ipv6/ranges.json b/discovery/linode/testdata/no_region_filter/v4/networking/ipv6/ranges.json new file mode 100644 index 0000000000..511a4d9a8c --- /dev/null +++ b/discovery/linode/testdata/no_region_filter/v4/networking/ipv6/ranges.json @@ -0,0 +1,19 @@ +{ + "data": [ + { + "range": "2600:3c03:e000:123::", + "prefix": 64, + "region": "us-east", + "route_target": "2600:3c03::f03c:92ff:fe1a:fb4c" + }, + { + "range": "2600:3c04:e001:456::", + "prefix": 64, + "region": "ca-central", + "route_target": "2600:3c04::f03c:92ff:fe1a:fb68" + } + ], + "page": 1, + "pages": 1, + "results": 2 +} diff --git a/discovery/linode/testdata/us-east/v4/account/events.json b/discovery/linode/testdata/us-east/v4/account/events.json new file mode 100644 index 0000000000..ca302e4fd0 --- /dev/null +++ b/discovery/linode/testdata/us-east/v4/account/events.json @@ -0,0 +1,6 @@ +{ + "data": [], + "results": 0, + "pages": 1, + "page": 1 +} diff --git a/discovery/linode/testdata/us-east/v4/linode/instances.json b/discovery/linode/testdata/us-east/v4/linode/instances.json new file mode 100644 index 0000000000..5e9a8f5abe --- /dev/null +++ b/discovery/linode/testdata/us-east/v4/linode/instances.json @@ -0,0 +1,97 @@ +{ + "data": [ + { + "id": 26838044, + "label": "prometheus-linode-sd-exporter-1", + "group": "", + "status": "running", + "created": "2021-05-12T04:23:44", + "updated": "2021-05-12T04:23:44", + "type": "g6-standard-2", + "ipv4": [ + "45.33.82.151", + "96.126.108.16", + "192.168.170.51", + "192.168.201.25" + ], + "ipv6": "2600:3c03::f03c:92ff:fe1a:1382/128", + "image": "linode/arch", + "region": "us-east", + "specs": { + "disk": 81920, + "memory": 4096, + "vcpus": 2, + "gpus": 0, + "transfer": 4000 + }, + "alerts": { + "cpu": 180, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + }, + { + "id": 26837992, + "label": "prometheus-linode-sd-exporter-4", + "group": "", + "status": "running", + "created": "2021-05-12T04:22:06", + "updated": "2021-05-12T04:22:06", + "type": "g6-nanode-1", + "ipv4": [ + "66.228.47.103", + "172.104.18.104", + "192.168.148.94" + ], + "ipv6": "2600:3c03::f03c:92ff:fe1a:fb4c/128", + "image": "linode/ubuntu20.04", + "region": "us-east", + "specs": { + "disk": 25600, + "memory": 1024, + "vcpus": 1, + "gpus": 0, + "transfer": 1000 + }, + "alerts": { + "cpu": 90, + "network_in": 10, + "network_out": 10, + "transfer_quota": 80, + "io": 10000 + }, + "backups": { + "enabled": false, + "schedule": { + "day": null, + "window": null + }, + "last_successful": null + }, + "hypervisor": "kvm", + "watchdog_enabled": true, + "tags": [ + "monitoring" + ] + } + ], + "page": 1, + "pages": 1, + "results": 2 +} + diff --git a/discovery/linode/testdata/us-east/v4/networking/ips.json b/discovery/linode/testdata/us-east/v4/networking/ips.json new file mode 100644 index 0000000000..388cf59659 --- /dev/null +++ b/discovery/linode/testdata/us-east/v4/networking/ips.json @@ -0,0 +1,106 @@ +{ + "page": 1, + "pages": 1, + "results": 9, + "data": [ + { + "address": "66.228.47.103", + "gateway": "66.228.47.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li328-103.members.linode.com", + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "172.104.18.104", + "gateway": "172.104.18.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li1832-104.members.linode.com", + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "192.168.148.94", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26837992, + "region": "us-east" + }, + { + "address": "192.168.170.51", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "96.126.108.16", + "gateway": "96.126.108.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li365-16.members.linode.com", + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "45.33.82.151", + "gateway": "45.33.82.1", + "subnet_mask": "255.255.255.0", + "prefix": 24, + "type": "ipv4", + "public": true, + "rdns": "li1028-151.members.linode.com", + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "192.168.201.25", + "gateway": null, + "subnet_mask": "255.255.128.0", + "prefix": 17, + "type": "ipv4", + "public": false, + "rdns": null, + "linode_id": 26838044, + "region": "us-east" + }, + { + "address": "2600:3c03::f03c:92ff:fe1a:fb4c", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26837992, + "region": "us-east", + "public": true + }, + { + "address": "2600:3c03::f03c:92ff:fe1a:1382", + "gateway": "fe80::1", + "subnet_mask": "ffff:ffff:ffff:ffff::", + "prefix": 64, + "type": "ipv6", + "rdns": null, + "linode_id": 26838044, + "region": "us-east", + "public": true + } + ] +} diff --git a/discovery/linode/testdata/us-east/v4/networking/ipv6/ranges.json b/discovery/linode/testdata/us-east/v4/networking/ipv6/ranges.json new file mode 100644 index 0000000000..34b2ae1cdd --- /dev/null +++ b/discovery/linode/testdata/us-east/v4/networking/ipv6/ranges.json @@ -0,0 +1,13 @@ +{ + "data": [ + { + "range": "2600:3c03:e000:123::", + "prefix": 64, + "region": "us-east", + "route_target": "2600:3c03::f03c:92ff:fe1a:fb4c" + } + ], + "page": 1, + "pages": 1, + "results": 1 +} diff --git a/discovery/manager.go b/discovery/manager.go index e3a2635575..f14071af30 100644 --- a/discovery/manager.go +++ b/discovery/manager.go @@ -169,6 +169,13 @@ func (m *Manager) Providers() []*Provider { return m.providers } +// UnregisterMetrics unregisters manager metrics. It does not unregister +// service discovery or refresh metrics, whose lifecycle is managed independent +// of the discovery Manager. +func (m *Manager) UnregisterMetrics() { + m.metrics.Unregister(m.registerer) +} + // Run starts the background processing. func (m *Manager) Run() error { go m.sender() diff --git a/discovery/manager_test.go b/discovery/manager_test.go index 52159d94f6..656d7c3c66 100644 --- a/discovery/manager_test.go +++ b/discovery/manager_test.go @@ -36,11 +36,11 @@ func TestMain(m *testing.M) { testutil.TolerantVerifyLeak(m) } -func NewTestMetrics(t *testing.T, reg prometheus.Registerer) (*RefreshMetricsManager, map[string]DiscovererMetrics) { +func NewTestMetrics(t *testing.T, reg prometheus.Registerer) (RefreshMetricsManager, map[string]DiscovererMetrics) { refreshMetrics := NewRefreshMetrics(reg) sdMetrics, err := RegisterSDMetrics(reg, refreshMetrics) require.NoError(t, err) - return &refreshMetrics, sdMetrics + return refreshMetrics, sdMetrics } // TestTargetUpdatesOrder checks that the target updates are received in the expected order. @@ -733,7 +733,6 @@ func verifySyncedPresence(t *testing.T, tGroups map[string][]*targetgroup.Group, t.Helper() if _, ok := tGroups[key]; !ok { t.Fatalf("'%s' should be present in Group map keys: %v", key, tGroups) - return } match := false var mergedTargets string @@ -1542,3 +1541,24 @@ func (t *testDiscoverer) update(tgs []*targetgroup.Group) { <-t.ready t.up <- tgs } + +func TestUnregisterMetrics(t *testing.T) { + reg := prometheus.NewRegistry() + // Check that all metrics can be unregistered, allowing a second manager to be created. + for i := 0; i < 2; i++ { + ctx, cancel := context.WithCancel(context.Background()) + + refreshMetrics, sdMetrics := NewTestMetrics(t, reg) + + discoveryManager := NewManager(ctx, log.NewNopLogger(), reg, sdMetrics) + // discoveryManager will be nil if there was an error configuring metrics. + require.NotNil(t, discoveryManager) + // Unregister all metrics. + discoveryManager.UnregisterMetrics() + for _, sdMetric := range sdMetrics { + sdMetric.Unregister() + } + refreshMetrics.Unregister() + cancel() + } +} diff --git a/discovery/marathon/marathon.go b/discovery/marathon/marathon.go index ecad108e4a..f833af47e7 100644 --- a/discovery/marathon/marathon.go +++ b/discovery/marathon/marathon.go @@ -339,7 +339,7 @@ type appListClient func(ctx context.Context, client *http.Client, url string) (* // fetchApps requests a list of applications from a marathon server. func fetchApps(ctx context.Context, client *http.Client, url string) (*appList, error) { - request, err := http.NewRequest("GET", url, nil) + request, err := http.NewRequest(http.MethodGet, url, nil) if err != nil { return nil, err } diff --git a/discovery/metrics.go b/discovery/metrics.go index a77b86d274..e738331a18 100644 --- a/discovery/metrics.go +++ b/discovery/metrics.go @@ -99,3 +99,12 @@ func NewManagerMetrics(registerer prometheus.Registerer, sdManagerName string) ( return m, nil } + +// Unregister unregisters all metrics. +func (m *Metrics) Unregister(registerer prometheus.Registerer) { + registerer.Unregister(m.FailedConfigs) + registerer.Unregister(m.DiscoveredTargets) + registerer.Unregister(m.ReceivedUpdates) + registerer.Unregister(m.DelayedUpdates) + registerer.Unregister(m.SentUpdates) +} diff --git a/discovery/openstack/mock_test.go b/discovery/openstack/mock_test.go index 4aa871e11f..b1267db90e 100644 --- a/discovery/openstack/mock_test.go +++ b/discovery/openstack/mock_test.go @@ -239,7 +239,7 @@ const hypervisorListBody = ` // HandleHypervisorListSuccessfully mocks os-hypervisors detail call. func (m *SDMock) HandleHypervisorListSuccessfully() { m.Mux.HandleFunc("/os-hypervisors/detail", func(w http.ResponseWriter, r *http.Request) { - testMethod(m.t, r, "GET") + testMethod(m.t, r, http.MethodGet) testHeader(m.t, r, "X-Auth-Token", tokenID) w.Header().Add("Content-Type", "application/json") @@ -536,7 +536,7 @@ const serverListBody = ` // HandleServerListSuccessfully mocks server detail call. func (m *SDMock) HandleServerListSuccessfully() { m.Mux.HandleFunc("/servers/detail", func(w http.ResponseWriter, r *http.Request) { - testMethod(m.t, r, "GET") + testMethod(m.t, r, http.MethodGet) testHeader(m.t, r, "X-Auth-Token", tokenID) w.Header().Add("Content-Type", "application/json") @@ -575,7 +575,7 @@ const listOutput = ` // HandleFloatingIPListSuccessfully mocks floating ips call. func (m *SDMock) HandleFloatingIPListSuccessfully() { m.Mux.HandleFunc("/os-floating-ips", func(w http.ResponseWriter, r *http.Request) { - testMethod(m.t, r, "GET") + testMethod(m.t, r, http.MethodGet) testHeader(m.t, r, "X-Auth-Token", tokenID) w.Header().Add("Content-Type", "application/json") diff --git a/discovery/puppetdb/puppetdb.go b/discovery/puppetdb/puppetdb.go index 3f9ad1f113..8c9ccde0a4 100644 --- a/discovery/puppetdb/puppetdb.go +++ b/discovery/puppetdb/puppetdb.go @@ -189,7 +189,7 @@ func (d *Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error) { return nil, err } - req, err := http.NewRequest("POST", d.url, bytes.NewBuffer(bodyBytes)) + req, err := http.NewRequest(http.MethodPost, d.url, bytes.NewBuffer(bodyBytes)) if err != nil { return nil, err } diff --git a/discovery/triton/triton.go b/discovery/triton/triton.go index e56b7951b6..675149f2a3 100644 --- a/discovery/triton/triton.go +++ b/discovery/triton/triton.go @@ -211,7 +211,7 @@ func (d *Discovery) refresh(ctx context.Context) ([]*targetgroup.Group, error) { endpoint = fmt.Sprintf("%s?groups=%s", endpoint, groups) } - req, err := http.NewRequest("GET", endpoint, nil) + req, err := http.NewRequest(http.MethodGet, endpoint, nil) if err != nil { return nil, err } diff --git a/discovery/xds/client.go b/discovery/xds/client.go index 9844c6d7ed..027ceb2715 100644 --- a/discovery/xds/client.go +++ b/discovery/xds/client.go @@ -179,7 +179,7 @@ func (rc *HTTPResourceClient) Fetch(ctx context.Context) (*v3.DiscoveryResponse, return nil, err } - request, err := http.NewRequest("POST", rc.endpoint, bytes.NewBuffer(reqBody)) + request, err := http.NewRequest(http.MethodPost, rc.endpoint, bytes.NewBuffer(reqBody)) if err != nil { return nil, err } diff --git a/docs/command-line/prometheus.md b/docs/command-line/prometheus.md index 2faea5b15e..93eaf251d0 100644 --- a/docs/command-line/prometheus.md +++ b/docs/command-line/prometheus.md @@ -54,7 +54,7 @@ The Prometheus monitoring server | --query.timeout | Maximum time a query may take before being aborted. Use with server mode only. | `2m` | | --query.max-concurrency | Maximum number of queries executed concurrently. Use with server mode only. | `20` | | --query.max-samples | Maximum number of samples a single query can load into memory. Note that queries will fail if they try to load more samples than this into memory, so this also limits the number of samples a query can return. Use with server mode only. | `50000000` | -| --enable-feature | Comma separated feature names to enable. Valid options: agent, auto-gomemlimit, exemplar-storage, expand-external-labels, memory-snapshot-on-shutdown, promql-at-modifier, promql-negative-offset, promql-per-step-stats, promql-experimental-functions, remote-write-receiver (DEPRECATED), extra-scrape-metrics, new-service-discovery-manager, auto-gomaxprocs, no-default-scrape-port, native-histograms, otlp-write-receiver. See https://prometheus.io/docs/prometheus/latest/feature_flags/ for more details. | | +| --enable-feature | Comma separated feature names to enable. Valid options: agent, auto-gomemlimit, exemplar-storage, expand-external-labels, memory-snapshot-on-shutdown, promql-per-step-stats, promql-experimental-functions, remote-write-receiver (DEPRECATED), extra-scrape-metrics, new-service-discovery-manager, auto-gomaxprocs, no-default-scrape-port, native-histograms, otlp-write-receiver, created-timestamp-zero-ingestion, concurrent-rule-eval. See https://prometheus.io/docs/prometheus/latest/feature_flags/ for more details. | | | --log.level | Only log messages with the given severity or above. One of: [debug, info, warn, error] | `info` | | --log.format | Output format of log messages. One of: [logfmt, json] | `logfmt` | diff --git a/docs/configuration/configuration.md b/docs/configuration/configuration.md index e134ae02b7..51eb84ae19 100644 --- a/docs/configuration/configuration.md +++ b/docs/configuration/configuration.md @@ -600,8 +600,10 @@ See below for the configuration options for Azure discovery: # The Azure environment. [ environment: | default = AzurePublicCloud ] -# The authentication method, either OAuth or ManagedIdentity. +# The authentication method, either OAuth, ManagedIdentity or SDK. # See https://docs.microsoft.com/en-us/azure/active-directory/managed-identities-azure-resources/overview +# SDK authentication method uses environment variables by default. +# See https://learn.microsoft.com/en-us/azure/developer/go/azure-sdk-authentication [ authentication_method: | default = OAuth] # The subscription ID. Always required. subscription_id: @@ -2447,11 +2449,15 @@ The following meta labels are available on targets during [relabeling](#relabel_ * `__meta_linode_private_ipv4`: the private IPv4 of the linode instance * `__meta_linode_public_ipv4`: the public IPv4 of the linode instance * `__meta_linode_public_ipv6`: the public IPv6 of the linode instance +* `__meta_linode_private_ipv4_rdns`: the reverse DNS for the first private IPv4 of the linode instance +* `__meta_linode_public_ipv4_rdns`: the reverse DNS for the first public IPv4 of the linode instance +* `__meta_linode_public_ipv6_rdns`: the reverse DNS for the first public IPv6 of the linode instance * `__meta_linode_region`: the region of the linode instance * `__meta_linode_type`: the type of the linode instance * `__meta_linode_status`: the status of the linode instance * `__meta_linode_tags`: a list of tags of the linode instance joined by the tag separator * `__meta_linode_group`: the display group a linode instance is a member of +* `__meta_linode_gpus`: the number of GPU's of the linode instance * `__meta_linode_hypervisor`: the virtualization software powering the linode instance * `__meta_linode_backups`: the backup service status of the linode instance * `__meta_linode_specs_disk_bytes`: the amount of storage space the linode instance has access to @@ -2459,6 +2465,7 @@ The following meta labels are available on targets during [relabeling](#relabel_ * `__meta_linode_specs_vcpus`: the number of VCPUS this linode has access to * `__meta_linode_specs_transfer_bytes`: the amount of network transfer the linode instance is allotted each month * `__meta_linode_extra_ips`: a list of all extra IPv4 addresses assigned to the linode instance joined by the tag separator +* `__meta_linode_ipv6_ranges`: a list of IPv6 ranges with mask assigned to the linode instance joined by the tag separator ```yaml # Authentication information used to authenticate to the API server. @@ -2489,6 +2496,9 @@ authorization: oauth2: [ ] +# Optional region to filter on. +[ region: ] + # Optional proxy URL. [ proxy_url: ] # Comma-separated string that can contain IPs, CIDR notation, domain names @@ -3226,7 +3236,7 @@ are set to the scheme and metrics path of the target respectively. The `__param_ label is set to the value of the first passed URL parameter called ``. The `__scrape_interval__` and `__scrape_timeout__` labels are set to the target's -interval and timeout. This is **experimental** and could change in the future. +interval and timeout. Additional labels prefixed with `__meta_` may be available during the relabeling phase. They are set by the service discovery mechanism that provided @@ -3619,6 +3629,11 @@ azuread: [ client_secret: ] [ tenant_id: ] ] + # Azure SDK auth. + # See https://learn.microsoft.com/en-us/azure/developer/go/azure-sdk-authentication + [ sdk: + [ tenant_id: ] ] + # Configures the remote write request's TLS settings. tls_config: [ ] diff --git a/documentation/examples/prometheus-linode.yml b/documentation/examples/prometheus-linode.yml index 993b6a5c12..fe1a740028 100644 --- a/documentation/examples/prometheus-linode.yml +++ b/documentation/examples/prometheus-linode.yml @@ -12,6 +12,7 @@ scrape_configs: linode_sd_configs: - authorization: credentials: "" + region: "us-east" relabel_configs: # Only scrape targets that have a tag 'monitoring'. - source_labels: [__meta_linode_tags] diff --git a/documentation/examples/remote_storage/go.mod b/documentation/examples/remote_storage/go.mod index 38497cbf50..917563f00c 100644 --- a/documentation/examples/remote_storage/go.mod +++ b/documentation/examples/remote_storage/go.mod @@ -9,7 +9,7 @@ require ( github.com/golang/snappy v0.0.4 github.com/influxdata/influxdb v1.11.5 github.com/prometheus/client_golang v1.19.0 - github.com/prometheus/common v0.49.0 + github.com/prometheus/common v0.50.0 github.com/prometheus/prometheus v0.50.1 github.com/stretchr/testify v1.9.0 ) @@ -58,17 +58,17 @@ require ( go.opentelemetry.io/otel/trace v1.22.0 // indirect go.uber.org/atomic v1.11.0 // indirect go.uber.org/multierr v1.11.0 // indirect - golang.org/x/crypto v0.19.0 // indirect + golang.org/x/crypto v0.21.0 // indirect golang.org/x/exp v0.0.0-20240119083558-1b970713d09a // indirect - golang.org/x/net v0.21.0 // indirect - golang.org/x/oauth2 v0.17.0 // indirect - golang.org/x/sys v0.17.0 // indirect + golang.org/x/net v0.22.0 // indirect + golang.org/x/oauth2 v0.18.0 // indirect + golang.org/x/sys v0.18.0 // indirect golang.org/x/text v0.14.0 // indirect golang.org/x/time v0.5.0 // indirect google.golang.org/appengine v1.6.8 // indirect google.golang.org/genproto/googleapis/rpc v0.0.0-20240116215550-a9fa1716bcac // indirect google.golang.org/grpc v1.61.0 // indirect - google.golang.org/protobuf v1.32.0 // indirect + google.golang.org/protobuf v1.33.0 // indirect gopkg.in/yaml.v2 v2.4.0 // indirect gopkg.in/yaml.v3 v3.0.1 // indirect k8s.io/apimachinery v0.28.6 // indirect diff --git a/documentation/examples/remote_storage/go.sum b/documentation/examples/remote_storage/go.sum index 6ffa445b29..50db8d7934 100644 --- a/documentation/examples/remote_storage/go.sum +++ b/documentation/examples/remote_storage/go.sum @@ -269,8 +269,8 @@ github.com/prometheus/common v0.4.1/go.mod h1:TNfzLD0ON7rHzMJeJkieUDPYmFC7Snx/y8 github.com/prometheus/common v0.10.0/go.mod h1:Tlit/dnDKsSWFlCLTWaA1cyBgKHSMdTB80sz/V91rCo= github.com/prometheus/common v0.26.0/go.mod h1:M7rCNAaPfAosfx8veZJCuw84e35h3Cfd9VFqTh1DIvc= github.com/prometheus/common v0.29.0/go.mod h1:vu+V0TpY+O6vW9J44gczi3Ap/oXXR10b+M/gUGO4Hls= -github.com/prometheus/common v0.49.0 h1:ToNTdK4zSnPVJmh698mGFkDor9wBI/iGaJy5dbH1EgI= -github.com/prometheus/common v0.49.0/go.mod h1:Kxm+EULxRbUkjGU6WFsQqo3ORzB4tyKvlWFOE9mB2sE= +github.com/prometheus/common v0.50.0 h1:YSZE6aa9+luNa2da6/Tik0q0A5AbR+U003TItK57CPQ= +github.com/prometheus/common v0.50.0/go.mod h1:wHFBCEVWVmHMUpg7pYcOm2QUR/ocQdYSJVQJKnHc3xQ= github.com/prometheus/common/sigv4 v0.1.0 h1:qoVebwtwwEhS85Czm2dSROY5fTo2PAPEVdDeppTwGX4= github.com/prometheus/common/sigv4 v0.1.0/go.mod h1:2Jkxxk9yYvCkE5G1sQT7GuEXm57JrvHu9k5YwTjsNtI= github.com/prometheus/procfs v0.0.0-20181005140218-185b4288413d/go.mod h1:c3At6R/oaqEKCNdg8wHV1ftS6bRYblBhIjjI8uT2IGk= @@ -332,8 +332,8 @@ golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACk golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI= golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto= golang.org/x/crypto v0.0.0-20210921155107-089bfa567519/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc= -golang.org/x/crypto v0.19.0 h1:ENy+Az/9Y1vSrlrvBSyna3PITt4tiZLf7sgCjZBX7Wo= -golang.org/x/crypto v0.19.0/go.mod h1:Iy9bg/ha4yyC70EfRS8jz+B6ybOBKMaSxLj6P6oBDfU= +golang.org/x/crypto v0.21.0 h1:X31++rzVUdKhX5sWmSOFZxx8UW/ldWx55cbf08iNAMA= +golang.org/x/crypto v0.21.0/go.mod h1:0BP7YvVV9gBbVKyeTG0Gyn+gZm94bibOW5BjDEYAOMs= golang.org/x/exp v0.0.0-20240119083558-1b970713d09a h1:Q8/wZp0KX97QFTc2ywcOE0YRjZPVIx+MXInMzdvQqcA= golang.org/x/exp v0.0.0-20240119083558-1b970713d09a/go.mod h1:idGWGoKP1toJGkd5/ig9ZLuPcZBC3ewk7SzmH0uou08= golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA= @@ -356,12 +356,12 @@ golang.org/x/net v0.0.0-20201110031124-69a78807bb2b/go.mod h1:sp8m0HH+o8qH0wwXwY golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg= golang.org/x/net v0.0.0-20210525063256-abc453219eb5/go.mod h1:9nx3DQGgdP8bBQD5qxJ1jj9UTztislL4KSBs9R2vV5Y= golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c= -golang.org/x/net v0.21.0 h1:AQyQV4dYCvJ7vGmJyKki9+PBdyvhkSd8EIx/qb0AYv4= -golang.org/x/net v0.21.0/go.mod h1:bIjVDfnllIU7BJ2DNgfnXvpSvtn8VRwhlsaeUTyUS44= +golang.org/x/net v0.22.0 h1:9sGLhx7iRIHEiX0oAJ3MRZMUCElJgy7Br1nO+AMN3Tc= +golang.org/x/net v0.22.0/go.mod h1:JKghWKKOSdJwpW2GEx0Ja7fmaKnMsbu+MWVZTokSYmg= golang.org/x/oauth2 v0.0.0-20190226205417-e64efc72b421/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw= golang.org/x/oauth2 v0.0.0-20210514164344-f6687ab2804c/go.mod h1:KelEdhl1UZF7XfJ4dDtk6s++YSgaE7mD/BuKKDLBl4A= -golang.org/x/oauth2 v0.17.0 h1:6m3ZPmLEFdVxKKWnKq4VqZ60gutO35zm+zrAHVmHyDQ= -golang.org/x/oauth2 v0.17.0/go.mod h1:OzPDGQiuQMguemayvdylqddI7qcD9lnSDb+1FiwQ5HA= +golang.org/x/oauth2 v0.18.0 h1:09qnuIAgzdx1XplqJvW6CQqMCtGZykZWcXzPMPUusvI= +golang.org/x/oauth2 v0.18.0/go.mod h1:Wf7knwG0MPoWIMMBgFlEaSUDaKskp0dCfrlJRJXbBi8= golang.org/x/sync v0.0.0-20181108010431-42b317875d0f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM= @@ -389,12 +389,12 @@ golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBc golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg= -golang.org/x/sys v0.17.0 h1:25cE3gD+tdBA7lp7QfhuV+rJiE9YXTcS3VG1SqssI/Y= -golang.org/x/sys v0.17.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA= +golang.org/x/sys v0.18.0 h1:DBdB3niSjOA/O0blCZBqDefyWNYveAYMNF1Wum0DYQ4= +golang.org/x/sys v0.18.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA= golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo= golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8= -golang.org/x/term v0.17.0 h1:mkTF7LCd6WGJNL3K1Ad7kwxNfYAW6a8a8QqtMblp/4U= -golang.org/x/term v0.17.0/go.mod h1:lLRBjIVuehSbZlaOtGMbcMncT+aqLLLmKrsjNrUguwk= +golang.org/x/term v0.18.0 h1:FcHjZXDMxI8mM3nwhX9HlKop4C0YQvCVCdwYl2wOtE8= +golang.org/x/term v0.18.0/go.mod h1:ILwASektA3OnRv7amZ1xhE/KTR+u50pbXfZ03+6Nx58= golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ= golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk= golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ= @@ -436,8 +436,8 @@ google.golang.org/protobuf v1.21.0/go.mod h1:47Nbq4nVaFHyn7ilMalzfO3qCViNmqZ2kzi google.golang.org/protobuf v1.23.0/go.mod h1:EGpADcykh3NcUnDUJcl1+ZksZNG86OlYog2l/sGQquU= google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw= google.golang.org/protobuf v1.26.0/go.mod h1:9q0QmTI4eRPtz6boOQmLYwt+qCgq0jsYwAQnmE0givc= -google.golang.org/protobuf v1.32.0 h1:pPC6BG5ex8PDFnkbrGU3EixyhKcQ2aDuBS36lqK/C7I= -google.golang.org/protobuf v1.32.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos= +google.golang.org/protobuf v1.33.0 h1:uNO2rsAINq/JlFpSdYEKIZ0uKD/R9cpdv0T+yoGwGmI= +google.golang.org/protobuf v1.33.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos= gopkg.in/alecthomas/kingpin.v2 v2.2.6/go.mod h1:FMv+mEhP44yOT+4EoQTLFTRgOQ1FBLkstjWtayDeSgw= gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0= gopkg.in/check.v1 v1.0.0-20190902080502-41f04d3bba15/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0= diff --git a/documentation/examples/remote_storage/remote_storage_adapter/influxdb/client_test.go b/documentation/examples/remote_storage/remote_storage_adapter/influxdb/client_test.go index cb56514e4b..a738c01dcd 100644 --- a/documentation/examples/remote_storage/remote_storage_adapter/influxdb/client_test.go +++ b/documentation/examples/remote_storage/remote_storage_adapter/influxdb/client_test.go @@ -74,7 +74,7 @@ testmetric,test_label=test_label_value2 value=5.1234 123456789123 server := httptest.NewServer(http.HandlerFunc( func(w http.ResponseWriter, r *http.Request) { - require.Equal(t, "POST", r.Method, "Unexpected method.") + require.Equal(t, http.MethodPost, r.Method, "Unexpected method.") require.Equal(t, "/write", r.URL.Path, "Unexpected path.") b, err := io.ReadAll(r.Body) require.NoError(t, err, "Error reading body.") diff --git a/documentation/examples/remote_storage/remote_storage_adapter/opentsdb/client.go b/documentation/examples/remote_storage/remote_storage_adapter/opentsdb/client.go index 0fa7c5a4b7..abb1d0b7d3 100644 --- a/documentation/examples/remote_storage/remote_storage_adapter/opentsdb/client.go +++ b/documentation/examples/remote_storage/remote_storage_adapter/opentsdb/client.go @@ -105,7 +105,7 @@ func (c *Client) Write(samples model.Samples) error { ctx, cancel := context.WithTimeout(context.Background(), c.timeout) defer cancel() - req, err := http.NewRequest("POST", u.String(), bytes.NewBuffer(buf)) + req, err := http.NewRequest(http.MethodPost, u.String(), bytes.NewBuffer(buf)) if err != nil { return err } diff --git a/go.mod b/go.mod index 8039a16af4..8e9824b49d 100644 --- a/go.mod +++ b/go.mod @@ -41,7 +41,7 @@ require ( github.com/json-iterator/go v1.1.12 github.com/klauspost/compress v1.17.7 github.com/kolo/xmlrpc v0.0.0-20220921171641-a4b6fa1dd06b - github.com/linode/linodego v1.29.0 + github.com/linode/linodego v1.30.0 github.com/miekg/dns v1.1.58 github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 github.com/mwitkow/go-conntrack v0.0.0-20190716064945-2f068394615f @@ -60,9 +60,9 @@ require ( github.com/shurcooL/httpfs v0.0.0-20230704072500-f1e31cf0ba5c github.com/stretchr/testify v1.9.0 github.com/vultr/govultr/v2 v2.17.2 - go.opentelemetry.io/collector/featuregate v1.3.0 - go.opentelemetry.io/collector/pdata v1.3.0 - go.opentelemetry.io/collector/semconv v0.96.0 + go.opentelemetry.io/collector/featuregate v1.4.0 + go.opentelemetry.io/collector/pdata v1.4.0 + go.opentelemetry.io/collector/semconv v0.97.0 go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 go.opentelemetry.io/otel v1.24.0 go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.24.0 @@ -74,7 +74,6 @@ require ( go.uber.org/automaxprocs v1.5.3 go.uber.org/goleak v1.3.0 go.uber.org/multierr v1.11.0 - golang.org/x/exp v0.0.0-20240119083558-1b970713d09a // indirect golang.org/x/net v0.22.0 golang.org/x/oauth2 v0.18.0 golang.org/x/sync v0.6.0 @@ -84,12 +83,12 @@ require ( google.golang.org/api v0.168.0 google.golang.org/genproto/googleapis/api v0.0.0-20240304212257-790db918fca8 google.golang.org/grpc v1.62.1 - google.golang.org/protobuf v1.32.0 + google.golang.org/protobuf v1.33.0 gopkg.in/yaml.v2 v2.4.0 gopkg.in/yaml.v3 v3.0.1 - k8s.io/api v0.29.2 - k8s.io/apimachinery v0.29.2 - k8s.io/client-go v0.29.2 + k8s.io/api v0.29.3 + k8s.io/apimachinery v0.29.3 + k8s.io/client-go v0.29.3 k8s.io/klog v1.0.0 k8s.io/klog/v2 v2.120.1 ) @@ -134,7 +133,7 @@ require ( github.com/golang-jwt/jwt/v5 v5.2.0 // indirect github.com/golang/glog v1.2.0 // indirect github.com/golang/groupcache v0.0.0-20210331224755-41bb18bfe9da // indirect - github.com/golang/protobuf v1.5.3 // indirect + github.com/golang/protobuf v1.5.4 // indirect github.com/google/gnostic-models v0.6.8 // indirect github.com/google/go-querystring v1.1.0 // indirect github.com/google/gofuzz v1.2.0 // indirect @@ -186,6 +185,7 @@ require ( go.opentelemetry.io/otel/metric v1.24.0 // indirect go.opentelemetry.io/proto/otlp v1.1.0 // indirect golang.org/x/crypto v0.21.0 // indirect + golang.org/x/exp v0.0.0-20240119083558-1b970713d09a // indirect golang.org/x/mod v0.16.0 // indirect golang.org/x/term v0.18.0 // indirect golang.org/x/text v0.14.0 // indirect diff --git a/go.sum b/go.sum index 599f22a1f7..135e3afafc 100644 --- a/go.sum +++ b/go.sum @@ -280,8 +280,8 @@ github.com/golang/protobuf v1.4.2/go.mod h1:oDoupMAO8OvCJWAcko0GGGIgR6R6ocIYbsSw github.com/golang/protobuf v1.4.3/go.mod h1:oDoupMAO8OvCJWAcko0GGGIgR6R6ocIYbsSw735rRwI= github.com/golang/protobuf v1.5.0/go.mod h1:FsONVRAS9T7sI+LIUmWTfcYkHO4aIWwzhcaSAoJOfIk= github.com/golang/protobuf v1.5.2/go.mod h1:XVQd3VNwM+JqD3oG2Ue2ip4fOMUkwXdXDdiuN0vRsmY= -github.com/golang/protobuf v1.5.3 h1:KhyjKVUg7Usr/dYsdSqoFveMYd5ko72D+zANwlG1mmg= -github.com/golang/protobuf v1.5.3/go.mod h1:XVQd3VNwM+JqD3oG2Ue2ip4fOMUkwXdXDdiuN0vRsmY= +github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek= +github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps= github.com/golang/snappy v0.0.0-20180518054509-2e65f85255db/go.mod h1:/XxbfmMg8lxefKM7IXC3fBNl/7bRcc72aCRzEWrmP2Q= github.com/golang/snappy v0.0.4 h1:yAGX7huGHXlcLOEtBnF4w7FQwA26wojNCwOYAEhLjQM= github.com/golang/snappy v0.0.4/go.mod h1:/XxbfmMg8lxefKM7IXC3fBNl/7bRcc72aCRzEWrmP2Q= @@ -424,8 +424,8 @@ github.com/inconshreveable/mousetrap v1.0.0/go.mod h1:PxqpIevigyE2G7u3NXJIT2ANyt github.com/influxdata/influxdb1-client v0.0.0-20191209144304-8bf82d3c094d/go.mod h1:qj24IKcXYK6Iy9ceXlo3Tc+vtHo9lIhSX5JddghvEPo= github.com/ionos-cloud/sdk-go/v6 v6.1.11 h1:J/uRN4UWO3wCyGOeDdMKv8LWRzKu6UIkLEaes38Kzh8= github.com/ionos-cloud/sdk-go/v6 v6.1.11/go.mod h1:EzEgRIDxBELvfoa/uBN0kOQaqovLjUWEB7iW4/Q+t4k= -github.com/jarcoal/httpmock v1.3.0 h1:2RJ8GP0IIaWwcC9Fp2BmVi8Kog3v2Hn7VXM3fTd+nuc= -github.com/jarcoal/httpmock v1.3.0/go.mod h1:3yb8rc4BI7TCBhFY8ng0gjuLKJNquuDNiPaZjnENuYg= +github.com/jarcoal/httpmock v1.3.1 h1:iUx3whfZWVf3jT01hQTO/Eo5sAYtB2/rqaUuOtpInww= +github.com/jarcoal/httpmock v1.3.1/go.mod h1:3yb8rc4BI7TCBhFY8ng0gjuLKJNquuDNiPaZjnENuYg= github.com/jmespath/go-jmespath v0.0.0-20180206201540-c2b33e8439af/go.mod h1:Nht3zPeWKUH0NzdCt2Blrr5ys8VGpn0CEB0cQHVjt7k= github.com/jmespath/go-jmespath v0.4.0 h1:BEgLn5cpjn8UN1mAw4NjwDrS35OdebyEtFe+9YPoQUg= github.com/jmespath/go-jmespath v0.4.0/go.mod h1:T8mJZnbsbmF+m6zOOFylbeCJqk5+pHWvzYPziyZiYoo= @@ -471,8 +471,8 @@ github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0 github.com/kylelemons/godebug v1.1.0/go.mod h1:9/0rRGxNHcop5bhtWyNeEfOS8JIWk580+fNqagV/RAw= github.com/lightstep/lightstep-tracer-common/golang/gogo v0.0.0-20190605223551-bc2310a04743/go.mod h1:qklhhLq1aX+mtWk9cPHPzaBjWImj5ULL6C7HFJtXQMM= github.com/lightstep/lightstep-tracer-go v0.18.1/go.mod h1:jlF1pusYV4pidLvZ+XD0UBX0ZE6WURAspgAczcDHrL4= -github.com/linode/linodego v1.29.0 h1:gDSQWAbKMAQX8db9FDCXHhodQPrJmLcmthjx6m+PyV4= -github.com/linode/linodego v1.29.0/go.mod h1:3k6WvCM10gillgYcnoLqIL23ST27BD9HhMsCJWb3Bpk= +github.com/linode/linodego v1.30.0 h1:6HJli+LX7NGu+Sne2G+ux790EkVOWOV/SR4mK3jcs6k= +github.com/linode/linodego v1.30.0/go.mod h1:/46h/XpmWi//oSA92GX2p3FIxb8HbX7grslPPQalR2o= github.com/lyft/protoc-gen-validate v0.0.13/go.mod h1:XbGvPuh87YZc5TdIa2/I4pLk0QoUACkjt2znoq26NVQ= github.com/mailru/easyjson v0.7.7 h1:UGYAvKxe3sBsEDzO8ZeWOSlIQfWFlxbzLZe7hwFURr0= github.com/mailru/easyjson v0.7.7/go.mod h1:xzfreul335JAWq5oZzymOObrkdz5UnU4kGfJJLY9Nlc= @@ -720,12 +720,12 @@ go.opencensus.io v0.22.3/go.mod h1:yxeiOL68Rb0Xd1ddK5vPZ/oVn4vY4Ynel7k9FzqtOIw= go.opencensus.io v0.22.4/go.mod h1:yxeiOL68Rb0Xd1ddK5vPZ/oVn4vY4Ynel7k9FzqtOIw= go.opencensus.io v0.24.0 h1:y73uSU6J157QMP2kn2r30vwW1A2W2WFwSCGnAVxeaD0= go.opencensus.io v0.24.0/go.mod h1:vNK8G9p7aAivkbmorf4v+7Hgx+Zs0yY+0fOtgBfjQKo= -go.opentelemetry.io/collector/featuregate v1.3.0 h1:nrFSx+zfjdisjE9oCx25Aep3nJ9RaUjeE1qFL6eovoU= -go.opentelemetry.io/collector/featuregate v1.3.0/go.mod h1:mm8+xyQfgDmqhyegZRNIQmoKsNnDTwWKFLsdMoXAb7A= -go.opentelemetry.io/collector/pdata v1.3.0 h1:JRYN7tVHYFwmtQhIYbxWeiKSa2L1nCohyAs8sYqKFZo= -go.opentelemetry.io/collector/pdata v1.3.0/go.mod h1:t7W0Undtes53HODPdSujPLTnfSR5fzT+WpL+RTaaayo= -go.opentelemetry.io/collector/semconv v0.96.0 h1:DrZy8BpzJDnN2zFxXRj6BhfGYxNlqpFHBqyuS9fVHRY= -go.opentelemetry.io/collector/semconv v0.96.0/go.mod h1:zOm/U3pgMIWcvrcnPbR9Xx2HinoXj46ERMK8PUV9wrs= +go.opentelemetry.io/collector/featuregate v1.4.0 h1:RWE9M659C9iuUQc4GzBsndkGHG1jIzIY+nZJWvcKy1M= +go.opentelemetry.io/collector/featuregate v1.4.0/go.mod h1:w7nUODKxEi3FLf1HslCiE6YWtMtOOrMnSwsDam8Mg9w= +go.opentelemetry.io/collector/pdata v1.4.0 h1:cA6Pr7Z2V7mE+i7FmYpavX7nefzd6H4CICgW0T9aJX0= +go.opentelemetry.io/collector/pdata v1.4.0/go.mod h1:0Ttp4wQinhV5oJTd9MjyvUegmZBO9O0nrlh/+EDLw+Q= +go.opentelemetry.io/collector/semconv v0.97.0 h1:iF3nTfThbiOwz7o5Pocn0dDnDoffd18ijDuf6Mwzi1s= +go.opentelemetry.io/collector/semconv v0.97.0/go.mod h1:8ElcRZ8Cdw5JnvhTOQOdYizkJaQ10Z2fS+R6djOnj6A= go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0 h1:jq9TW8u3so/bN+JPT166wjOI6/vQPF6Xe7nMNIltagk= go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.49.0/go.mod h1:p8pYQP+m5XfbZm9fxtSKAbM6oIllS7s2AfxrChvc7iw= go.opentelemetry.io/otel v1.24.0 h1:0LAOdjNmQeSTzGBzduGe/rU4tZhMwL5rWgtp9Ku5Jfo= @@ -1118,8 +1118,8 @@ google.golang.org/protobuf v1.24.0/go.mod h1:r/3tXBNzIEhYS9I1OUVjXDlt8tc493IdKGj google.golang.org/protobuf v1.25.0/go.mod h1:9JNX74DMeImyA3h4bdi1ymwjUzf21/xIlbajtzgsN7c= google.golang.org/protobuf v1.26.0-rc.1/go.mod h1:jlhhOSvTdKEhbULTjvd4ARK9grFBp09yW+WbY/TyQbw= google.golang.org/protobuf v1.26.0/go.mod h1:9q0QmTI4eRPtz6boOQmLYwt+qCgq0jsYwAQnmE0givc= -google.golang.org/protobuf v1.32.0 h1:pPC6BG5ex8PDFnkbrGU3EixyhKcQ2aDuBS36lqK/C7I= -google.golang.org/protobuf v1.32.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos= +google.golang.org/protobuf v1.33.0 h1:uNO2rsAINq/JlFpSdYEKIZ0uKD/R9cpdv0T+yoGwGmI= +google.golang.org/protobuf v1.33.0/go.mod h1:c6P6GXX6sHbq/GpV6MGZEdwhWPcYBgnhAHhKbcUYpos= gopkg.in/alecthomas/kingpin.v2 v2.2.6/go.mod h1:FMv+mEhP44yOT+4EoQTLFTRgOQ1FBLkstjWtayDeSgw= gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0= gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0= @@ -1161,12 +1161,12 @@ honnef.co/go/tools v0.0.0-20190523083050-ea95bdfd59fc/go.mod h1:rf3lG4BRIbNafJWh honnef.co/go/tools v0.0.1-2019.2.3/go.mod h1:a3bituU0lyd329TUQxRnasdCoJDkEUEAqEt0JzvZhAg= honnef.co/go/tools v0.0.1-2020.1.3/go.mod h1:X/FiERA/W4tHapMX5mGpAtMSVEeEUOyHaw9vFzvIQ3k= honnef.co/go/tools v0.0.1-2020.1.4/go.mod h1:X/FiERA/W4tHapMX5mGpAtMSVEeEUOyHaw9vFzvIQ3k= -k8s.io/api v0.29.2 h1:hBC7B9+MU+ptchxEqTNW2DkUosJpp1P+Wn6YncZ474A= -k8s.io/api v0.29.2/go.mod h1:sdIaaKuU7P44aoyyLlikSLayT6Vb7bvJNCX105xZXY0= -k8s.io/apimachinery v0.29.2 h1:EWGpfJ856oj11C52NRCHuU7rFDwxev48z+6DSlGNsV8= -k8s.io/apimachinery v0.29.2/go.mod h1:6HVkd1FwxIagpYrHSwJlQqZI3G9LfYWRPAkUvLnXTKU= -k8s.io/client-go v0.29.2 h1:FEg85el1TeZp+/vYJM7hkDlSTFZ+c5nnK44DJ4FyoRg= -k8s.io/client-go v0.29.2/go.mod h1:knlvFZE58VpqbQpJNbCbctTVXcd35mMyAAwBdpt4jrA= +k8s.io/api v0.29.3 h1:2ORfZ7+bGC3YJqGpV0KSDDEVf8hdGQ6A03/50vj8pmw= +k8s.io/api v0.29.3/go.mod h1:y2yg2NTyHUUkIoTC+phinTnEa3KFM6RZ3szxt014a80= +k8s.io/apimachinery v0.29.3 h1:2tbx+5L7RNvqJjn7RIuIKu9XTsIZ9Z5wX2G22XAa5EU= +k8s.io/apimachinery v0.29.3/go.mod h1:hx/S4V2PNW4OMg3WizRrHutyB5la0iCUbZym+W0EQIU= +k8s.io/client-go v0.29.3 h1:R/zaZbEAxqComZ9FHeQwOh3Y1ZUs7FaHKZdQtIc2WZg= +k8s.io/client-go v0.29.3/go.mod h1:tkDisCvgPfiRpxGnOORfkljmS+UrW+WtXAy2fTvXJB0= k8s.io/kube-openapi v0.0.0-20231010175941-2dd684a91f00 h1:aVUu9fTY98ivBPKR9Y5w/AuzbMm96cd3YHRTU83I780= k8s.io/kube-openapi v0.0.0-20231010175941-2dd684a91f00/go.mod h1:AsvuZPBlUDVuCdzJ87iajxtXuR9oktsTctW/R9wwouA= k8s.io/utils v0.0.0-20230726121419-3b25d923346b h1:sgn3ZU783SCgtaSJjpcVVlRqd6GSnlTLKgpAAttJvpI= diff --git a/model/labels/labels_common.go b/model/labels/labels_common.go index 4c4a87e872..f46321c97e 100644 --- a/model/labels/labels_common.go +++ b/model/labels/labels_common.go @@ -39,7 +39,8 @@ type Label struct { } func (ls Labels) String() string { - var b bytes.Buffer + var bytea [1024]byte // On stack to avoid memory allocation while building the output. + b := bytes.NewBuffer(bytea[:0]) b.WriteByte('{') i := 0 @@ -50,7 +51,7 @@ func (ls Labels) String() string { } b.WriteString(l.Name) b.WriteByte('=') - b.WriteString(strconv.Quote(l.Value)) + b.Write(strconv.AppendQuote(b.AvailableBuffer(), l.Value)) i++ }) b.WriteByte('}') diff --git a/model/labels/labels_stringlabels.go b/model/labels/labels_stringlabels.go index 2e718c2b1f..9ef764daec 100644 --- a/model/labels/labels_stringlabels.go +++ b/model/labels/labels_stringlabels.go @@ -363,13 +363,11 @@ func Compare(a, b Labels) int { // Now we know that there is some difference before the end of a and b. // Go back through the fields and find which field that difference is in. - firstCharDifferent := i - for i = 0; ; { - size, nextI := decodeSize(a.data, i) - if nextI+size > firstCharDifferent { - break - } + firstCharDifferent, i := i, 0 + size, nextI := decodeSize(a.data, i) + for nextI+size <= firstCharDifferent { i = nextI + size + size, nextI = decodeSize(a.data, i) } // Difference is inside this entry. aStr, _ := decodeString(a.data, i) diff --git a/model/labels/labels_test.go b/model/labels/labels_test.go index 359e6de3b8..3d6e7659f4 100644 --- a/model/labels/labels_test.go +++ b/model/labels/labels_test.go @@ -16,6 +16,7 @@ package labels import ( "encoding/json" "fmt" + "net/http" "strings" "testing" @@ -25,24 +26,31 @@ import ( func TestLabels_String(t *testing.T) { cases := []struct { - lables Labels + labels Labels expected string }{ { - lables: FromStrings("t1", "t1", "t2", "t2"), + labels: FromStrings("t1", "t1", "t2", "t2"), expected: "{t1=\"t1\", t2=\"t2\"}", }, { - lables: Labels{}, + labels: Labels{}, expected: "{}", }, } for _, c := range cases { - str := c.lables.String() + str := c.labels.String() require.Equal(t, c.expected, str) } } +func BenchmarkString(b *testing.B) { + ls := New(benchmarkLabels...) + for i := 0; i < b.N; i++ { + _ = ls.String() + } +} + func TestLabels_MatchLabels(t *testing.T) { labels := FromStrings( "__name__", "ALERTS", @@ -529,6 +537,16 @@ var comparisonBenchmarkScenarios = []struct { FromStrings("aaa", "bbb", "ccc", "ddd", "eee", "fff", "ggg", "hhh", "iii", "jjj", "kkk", "lll", "mmm", "nnn", "ooo", "ppp", "qqq", "rrz"), FromStrings("aaa", "bbb", "ccc", "ddd", "eee", "fff", "ggg", "hhh", "iii", "jjj", "kkk", "lll", "mmm", "nnn", "ooo", "ppp", "qqq", "rrr"), }, + { + "real long equal", + FromStrings("__name__", "kube_pod_container_status_last_terminated_exitcode", "cluster", "prod-af-north-0", " container", "prometheus", "instance", "kube-state-metrics-0:kube-state-metrics:ksm", "job", "kube-state-metrics/kube-state-metrics", " namespace", "observability-prometheus", "pod", "observability-prometheus-0", "uid", "d3ec90b2-4975-4607-b45d-b9ad64bb417e"), + FromStrings("__name__", "kube_pod_container_status_last_terminated_exitcode", "cluster", "prod-af-north-0", " container", "prometheus", "instance", "kube-state-metrics-0:kube-state-metrics:ksm", "job", "kube-state-metrics/kube-state-metrics", " namespace", "observability-prometheus", "pod", "observability-prometheus-0", "uid", "d3ec90b2-4975-4607-b45d-b9ad64bb417e"), + }, + { + "real long different end", + FromStrings("__name__", "kube_pod_container_status_last_terminated_exitcode", "cluster", "prod-af-north-0", " container", "prometheus", "instance", "kube-state-metrics-0:kube-state-metrics:ksm", "job", "kube-state-metrics/kube-state-metrics", " namespace", "observability-prometheus", "pod", "observability-prometheus-0", "uid", "d3ec90b2-4975-4607-b45d-b9ad64bb417e"), + FromStrings("__name__", "kube_pod_container_status_last_terminated_exitcode", "cluster", "prod-af-north-0", " container", "prometheus", "instance", "kube-state-metrics-0:kube-state-metrics:ksm", "job", "kube-state-metrics/kube-state-metrics", " namespace", "observability-prometheus", "pod", "observability-prometheus-0", "uid", "deadbeef-0000-1111-2222-b9ad64bb417e"), + }, } func BenchmarkLabels_Equals(b *testing.B) { @@ -789,24 +807,24 @@ func BenchmarkLabels_Hash(b *testing.B) { } } -func BenchmarkBuilder(b *testing.B) { - m := []Label{ - {"job", "node"}, - {"instance", "123.123.1.211:9090"}, - {"path", "/api/v1/namespaces//deployments/"}, - {"method", "GET"}, - {"namespace", "system"}, - {"status", "500"}, - {"prometheus", "prometheus-core-1"}, - {"datacenter", "eu-west-1"}, - {"pod_name", "abcdef-99999-defee"}, - } +var benchmarkLabels = []Label{ + {"job", "node"}, + {"instance", "123.123.1.211:9090"}, + {"path", "/api/v1/namespaces//deployments/"}, + {"method", http.MethodGet}, + {"namespace", "system"}, + {"status", "500"}, + {"prometheus", "prometheus-core-1"}, + {"datacenter", "eu-west-1"}, + {"pod_name", "abcdef-99999-defee"}, +} +func BenchmarkBuilder(b *testing.B) { var l Labels builder := NewBuilder(EmptyLabels()) for i := 0; i < b.N; i++ { builder.Reset(EmptyLabels()) - for _, l := range m { + for _, l := range benchmarkLabels { builder.Set(l.Name, l.Value) } l = builder.Labels() @@ -815,18 +833,7 @@ func BenchmarkBuilder(b *testing.B) { } func BenchmarkLabels_Copy(b *testing.B) { - m := map[string]string{ - "job": "node", - "instance": "123.123.1.211:9090", - "path": "/api/v1/namespaces//deployments/", - "method": "GET", - "namespace": "system", - "status": "500", - "prometheus": "prometheus-core-1", - "datacenter": "eu-west-1", - "pod_name": "abcdef-99999-defee", - } - l := FromMap(m) + l := New(benchmarkLabels...) for i := 0; i < b.N; i++ { l = l.Copy() diff --git a/model/labels/matcher.go b/model/labels/matcher.go index f299c40f64..1282f80d63 100644 --- a/model/labels/matcher.go +++ b/model/labels/matcher.go @@ -118,3 +118,30 @@ func (m *Matcher) GetRegexString() string { } return m.re.GetRegexString() } + +// SetMatches returns a set of equality matchers for the current regex matchers if possible. +// For examples the regexp `a(b|f)` will returns "ab" and "af". +// Returns nil if we can't replace the regexp by only equality matchers. +func (m *Matcher) SetMatches() []string { + if m.re == nil { + return nil + } + return m.re.SetMatches() +} + +// Prefix returns the required prefix of the value to match, if possible. +// It will be empty if it's an equality matcher or if the prefix can't be determined. +func (m *Matcher) Prefix() string { + if m.re == nil { + return "" + } + return m.re.prefix +} + +// IsRegexOptimized returns whether regex is optimized. +func (m *Matcher) IsRegexOptimized() bool { + if m.re == nil { + return false + } + return m.re.IsOptimized() +} diff --git a/model/labels/matcher_test.go b/model/labels/matcher_test.go index d26e9329f2..c23deafe61 100644 --- a/model/labels/matcher_test.go +++ b/model/labels/matcher_test.go @@ -14,13 +14,14 @@ package labels import ( + "fmt" "testing" "github.com/stretchr/testify/require" ) func mustNewMatcher(t *testing.T, mType MatchType, value string) *Matcher { - m, err := NewMatcher(mType, "", value) + m, err := NewMatcher(mType, "test_label_name", value) require.NoError(t, err) return m } @@ -81,6 +82,21 @@ func TestMatcher(t *testing.T) { value: "foo-bar", match: false, }, + { + matcher: mustNewMatcher(t, MatchRegexp, "$*bar"), + value: "foo-bar", + match: false, + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "bar^+"), + value: "foo-bar", + match: false, + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "$+bar"), + value: "foo-bar", + match: false, + }, } for _, test := range tests { @@ -118,6 +134,82 @@ func TestInverse(t *testing.T) { } } +func TestPrefix(t *testing.T) { + for i, tc := range []struct { + matcher *Matcher + prefix string + }{ + { + matcher: mustNewMatcher(t, MatchEqual, "abc"), + prefix: "", + }, + { + matcher: mustNewMatcher(t, MatchNotEqual, "abc"), + prefix: "", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abc.+"), + prefix: "abc", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abcd|abc.+"), + prefix: "abc", + }, + { + matcher: mustNewMatcher(t, MatchNotRegexp, "abcd|abc.+"), + prefix: "abc", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abc(def|ghj)|ab|a."), + prefix: "a", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "foo.+bar|foo.*baz"), + prefix: "foo", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abc|.*"), + prefix: "", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abc|def"), + prefix: "", + }, + { + matcher: mustNewMatcher(t, MatchRegexp, ".+def"), + prefix: "", + }, + } { + t.Run(fmt.Sprintf("%d: %s", i, tc.matcher), func(t *testing.T) { + require.Equal(t, tc.prefix, tc.matcher.Prefix()) + }) + } +} + +func TestIsRegexOptimized(t *testing.T) { + for i, tc := range []struct { + matcher *Matcher + isRegexOptimized bool + }{ + { + matcher: mustNewMatcher(t, MatchEqual, "abc"), + isRegexOptimized: false, + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "."), + isRegexOptimized: false, + }, + { + matcher: mustNewMatcher(t, MatchRegexp, "abc.+"), + isRegexOptimized: true, + }, + } { + t.Run(fmt.Sprintf("%d: %s", i, tc.matcher), func(t *testing.T) { + require.Equal(t, tc.isRegexOptimized, tc.matcher.IsRegexOptimized()) + }) + } +} + func BenchmarkMatchType_String(b *testing.B) { for i := 0; i <= b.N; i++ { _ = MatchType(i % int(MatchNotRegexp+1)).String() diff --git a/model/labels/regexp.go b/model/labels/regexp.go index 14319c7f7a..f35dc76f60 100644 --- a/model/labels/regexp.go +++ b/model/labels/regexp.go @@ -14,73 +14,348 @@ package labels import ( + "slices" "strings" "github.com/grafana/regexp" "github.com/grafana/regexp/syntax" ) -type FastRegexMatcher struct { - re *regexp.Regexp - prefix string - suffix string - contains string +const ( + maxSetMatches = 256 - // shortcut for literals - literal bool - value string + // The minimum number of alternate values a regex should have to trigger + // the optimization done by optimizeEqualStringMatchers() and so use a map + // to match values instead of iterating over a list. This value has + // been computed running BenchmarkOptimizeEqualStringMatchers. + minEqualMultiStringMatcherMapThreshold = 16 +) + +type FastRegexMatcher struct { + // Under some conditions, re is nil because the expression is never parsed. + // We store the original string to be able to return it in GetRegexString(). + reString string + re *regexp.Regexp + + setMatches []string + stringMatcher StringMatcher + prefix string + suffix string + contains string + + // matchString is the "compiled" function to run by MatchString(). + matchString func(string) bool } func NewFastRegexMatcher(v string) (*FastRegexMatcher, error) { - if isLiteral(v) { - return &FastRegexMatcher{literal: true, value: v}, nil - } - re, err := regexp.Compile("^(?:" + v + ")$") - if err != nil { - return nil, err - } - - parsed, err := syntax.Parse(v, syntax.Perl) - if err != nil { - return nil, err - } - m := &FastRegexMatcher{ - re: re, + reString: v, } - if parsed.Op == syntax.OpConcat { - m.prefix, m.suffix, m.contains = optimizeConcatRegex(parsed) + m.stringMatcher, m.setMatches = optimizeAlternatingLiterals(v) + if m.stringMatcher != nil { + // If we already have a string matcher, we don't need to parse the regex + // or compile the matchString function. This also avoids the behavior in + // compileMatchStringFunction where it prefers to use setMatches when + // available, even if the string matcher is faster. + m.matchString = m.stringMatcher.Matches + } else { + parsed, err := syntax.Parse(v, syntax.Perl) + if err != nil { + return nil, err + } + // Simplify the syntax tree to run faster. + parsed = parsed.Simplify() + m.re, err = regexp.Compile("^(?:" + parsed.String() + ")$") + if err != nil { + return nil, err + } + if parsed.Op == syntax.OpConcat { + m.prefix, m.suffix, m.contains = optimizeConcatRegex(parsed) + } + if matches, caseSensitive := findSetMatches(parsed); caseSensitive { + m.setMatches = matches + } + m.stringMatcher = stringMatcherFromRegexp(parsed) + m.matchString = m.compileMatchStringFunction() } return m, nil } +// compileMatchStringFunction returns the function to run by MatchString(). +func (m *FastRegexMatcher) compileMatchStringFunction() func(string) bool { + // If the only optimization available is the string matcher, then we can just run it. + if len(m.setMatches) == 0 && m.prefix == "" && m.suffix == "" && m.contains == "" && m.stringMatcher != nil { + return m.stringMatcher.Matches + } + + return func(s string) bool { + if len(m.setMatches) != 0 { + for _, match := range m.setMatches { + if match == s { + return true + } + } + return false + } + if m.prefix != "" && !strings.HasPrefix(s, m.prefix) { + return false + } + if m.suffix != "" && !strings.HasSuffix(s, m.suffix) { + return false + } + if m.contains != "" && !strings.Contains(s, m.contains) { + return false + } + if m.stringMatcher != nil { + return m.stringMatcher.Matches(s) + } + return m.re.MatchString(s) + } +} + +// IsOptimized returns true if any fast-path optimization is applied to the +// regex matcher. +func (m *FastRegexMatcher) IsOptimized() bool { + return len(m.setMatches) > 0 || m.stringMatcher != nil || m.prefix != "" || m.suffix != "" || m.contains != "" +} + +// findSetMatches extract equality matches from a regexp. +// Returns nil if we can't replace the regexp by only equality matchers or the regexp contains +// a mix of case sensitive and case insensitive matchers. +func findSetMatches(re *syntax.Regexp) (matches []string, caseSensitive bool) { + clearBeginEndText(re) + + return findSetMatchesInternal(re, "") +} + +func findSetMatchesInternal(re *syntax.Regexp, base string) (matches []string, caseSensitive bool) { + switch re.Op { + case syntax.OpBeginText: + // Correctly handling the begin text operator inside a regex is tricky, + // so in this case we fallback to the regex engine. + return nil, false + case syntax.OpEndText: + // Correctly handling the end text operator inside a regex is tricky, + // so in this case we fallback to the regex engine. + return nil, false + case syntax.OpLiteral: + return []string{base + string(re.Rune)}, isCaseSensitive(re) + case syntax.OpEmptyMatch: + if base != "" { + return []string{base}, isCaseSensitive(re) + } + case syntax.OpAlternate: + return findSetMatchesFromAlternate(re, base) + case syntax.OpCapture: + clearCapture(re) + return findSetMatchesInternal(re, base) + case syntax.OpConcat: + return findSetMatchesFromConcat(re, base) + case syntax.OpCharClass: + if len(re.Rune)%2 != 0 { + return nil, false + } + var matches []string + var totalSet int + for i := 0; i+1 < len(re.Rune); i += 2 { + totalSet += int(re.Rune[i+1]-re.Rune[i]) + 1 + } + // limits the total characters that can be used to create matches. + // In some case like negation [^0-9] a lot of possibilities exists and that + // can create thousands of possible matches at which points we're better off using regexp. + if totalSet > maxSetMatches { + return nil, false + } + for i := 0; i+1 < len(re.Rune); i += 2 { + lo, hi := re.Rune[i], re.Rune[i+1] + for c := lo; c <= hi; c++ { + matches = append(matches, base+string(c)) + } + } + return matches, isCaseSensitive(re) + default: + return nil, false + } + return nil, false +} + +func findSetMatchesFromConcat(re *syntax.Regexp, base string) (matches []string, matchesCaseSensitive bool) { + if len(re.Sub) == 0 { + return nil, false + } + clearCapture(re.Sub...) + + matches = []string{base} + + for i := 0; i < len(re.Sub); i++ { + var newMatches []string + for j, b := range matches { + m, caseSensitive := findSetMatchesInternal(re.Sub[i], b) + if m == nil { + return nil, false + } + if tooManyMatches(newMatches, m...) { + return nil, false + } + + // All matches must have the same case sensitivity. If it's the first set of matches + // returned, we store its sensitivity as the expected case, and then we'll check all + // other ones. + if i == 0 && j == 0 { + matchesCaseSensitive = caseSensitive + } + if matchesCaseSensitive != caseSensitive { + return nil, false + } + + newMatches = append(newMatches, m...) + } + matches = newMatches + } + + return matches, matchesCaseSensitive +} + +func findSetMatchesFromAlternate(re *syntax.Regexp, base string) (matches []string, matchesCaseSensitive bool) { + for i, sub := range re.Sub { + found, caseSensitive := findSetMatchesInternal(sub, base) + if found == nil { + return nil, false + } + if tooManyMatches(matches, found...) { + return nil, false + } + + // All matches must have the same case sensitivity. If it's the first set of matches + // returned, we store its sensitivity as the expected case, and then we'll check all + // other ones. + if i == 0 { + matchesCaseSensitive = caseSensitive + } + if matchesCaseSensitive != caseSensitive { + return nil, false + } + + matches = append(matches, found...) + } + + return matches, matchesCaseSensitive +} + +// clearCapture removes capture operation as they are not used for matching. +func clearCapture(regs ...*syntax.Regexp) { + for _, r := range regs { + // Iterate on the regexp because capture groups could be nested. + for r.Op == syntax.OpCapture { + *r = *r.Sub[0] + } + } +} + +// clearBeginEndText removes the begin and end text from the regexp. Prometheus regexp are anchored to the beginning and end of the string. +func clearBeginEndText(re *syntax.Regexp) { + // Do not clear begin/end text from an alternate operator because it could + // change the actual regexp properties. + if re.Op == syntax.OpAlternate { + return + } + + if len(re.Sub) == 0 { + return + } + if len(re.Sub) == 1 { + if re.Sub[0].Op == syntax.OpBeginText || re.Sub[0].Op == syntax.OpEndText { + // We need to remove this element. Since it's the only one, we convert into a matcher of an empty string. + // OpEmptyMatch is regexp's nop operator. + re.Op = syntax.OpEmptyMatch + re.Sub = nil + return + } + } + if re.Sub[0].Op == syntax.OpBeginText { + re.Sub = re.Sub[1:] + } + if re.Sub[len(re.Sub)-1].Op == syntax.OpEndText { + re.Sub = re.Sub[:len(re.Sub)-1] + } +} + +// isCaseInsensitive tells if a regexp is case insensitive. +// The flag should be check at each level of the syntax tree. +func isCaseInsensitive(reg *syntax.Regexp) bool { + return (reg.Flags & syntax.FoldCase) != 0 +} + +// isCaseSensitive tells if a regexp is case sensitive. +// The flag should be check at each level of the syntax tree. +func isCaseSensitive(reg *syntax.Regexp) bool { + return !isCaseInsensitive(reg) +} + +// tooManyMatches guards against creating too many set matches. +func tooManyMatches(matches []string, added ...string) bool { + return len(matches)+len(added) > maxSetMatches +} + func (m *FastRegexMatcher) MatchString(s string) bool { - if m.literal { - return s == m.value - } - if m.prefix != "" && !strings.HasPrefix(s, m.prefix) { - return false - } - if m.suffix != "" && !strings.HasSuffix(s, m.suffix) { - return false - } - if m.contains != "" && !strings.Contains(s, m.contains) { - return false - } - return m.re.MatchString(s) + return m.matchString(s) +} + +func (m *FastRegexMatcher) SetMatches() []string { + // IMPORTANT: always return a copy, otherwise if the caller manipulate this slice it will + // also get manipulated in the cached FastRegexMatcher instance. + return slices.Clone(m.setMatches) } func (m *FastRegexMatcher) GetRegexString() string { - if m.literal { - return m.value - } - return m.re.String() + return m.reString } -func isLiteral(re string) bool { - return regexp.QuoteMeta(re) == re +// optimizeAlternatingLiterals optimizes a regex of the form +// +// `literal1|literal2|literal3|...` +// +// this function returns an optimized StringMatcher or nil if the regex +// cannot be optimized in this way, and a list of setMatches up to maxSetMatches. +func optimizeAlternatingLiterals(s string) (StringMatcher, []string) { + if len(s) == 0 { + return emptyStringMatcher{}, nil + } + + estimatedAlternates := strings.Count(s, "|") + 1 + + // If there are no alternates, check if the string is a literal + if estimatedAlternates == 1 { + if regexp.QuoteMeta(s) == s { + return &equalStringMatcher{s: s, caseSensitive: true}, []string{s} + } + return nil, nil + } + + multiMatcher := newEqualMultiStringMatcher(true, estimatedAlternates) + + for end := strings.IndexByte(s, '|'); end > -1; end = strings.IndexByte(s, '|') { + // Split the string into the next literal and the remainder + subMatch := s[:end] + s = s[end+1:] + + // break if any of the submatches are not literals + if regexp.QuoteMeta(subMatch) != subMatch { + return nil, nil + } + + multiMatcher.add(subMatch) + } + + // break if the remainder is not a literal + if regexp.QuoteMeta(s) != s { + return nil, nil + } + multiMatcher.add(s) + + return multiMatcher, multiMatcher.setMatches() } // optimizeConcatRegex returns literal prefix/suffix text that can be safely @@ -123,3 +398,540 @@ func optimizeConcatRegex(r *syntax.Regexp) (prefix, suffix, contains string) { return } + +// StringMatcher is a matcher that matches a string in place of a regular expression. +type StringMatcher interface { + Matches(s string) bool +} + +// stringMatcherFromRegexp attempts to replace a common regexp with a string matcher. +// It returns nil if the regexp is not supported. +func stringMatcherFromRegexp(re *syntax.Regexp) StringMatcher { + clearBeginEndText(re) + + m := stringMatcherFromRegexpInternal(re) + m = optimizeEqualStringMatchers(m, minEqualMultiStringMatcherMapThreshold) + + return m +} + +func stringMatcherFromRegexpInternal(re *syntax.Regexp) StringMatcher { + clearCapture(re) + + switch re.Op { + case syntax.OpBeginText: + // Correctly handling the begin text operator inside a regex is tricky, + // so in this case we fallback to the regex engine. + return nil + case syntax.OpEndText: + // Correctly handling the end text operator inside a regex is tricky, + // so in this case we fallback to the regex engine. + return nil + case syntax.OpPlus: + if re.Sub[0].Op != syntax.OpAnyChar && re.Sub[0].Op != syntax.OpAnyCharNotNL { + return nil + } + return &anyNonEmptyStringMatcher{ + matchNL: re.Sub[0].Op == syntax.OpAnyChar, + } + case syntax.OpStar: + if re.Sub[0].Op != syntax.OpAnyChar && re.Sub[0].Op != syntax.OpAnyCharNotNL { + return nil + } + + // If the newline is valid, than this matcher literally match any string (even empty). + if re.Sub[0].Op == syntax.OpAnyChar { + return trueMatcher{} + } + + // Any string is fine (including an empty one), as far as it doesn't contain any newline. + return anyStringWithoutNewlineMatcher{} + case syntax.OpQuest: + // Only optimize for ".?". + if len(re.Sub) != 1 || (re.Sub[0].Op != syntax.OpAnyChar && re.Sub[0].Op != syntax.OpAnyCharNotNL) { + return nil + } + + return &zeroOrOneCharacterStringMatcher{ + matchNL: re.Sub[0].Op == syntax.OpAnyChar, + } + case syntax.OpEmptyMatch: + return emptyStringMatcher{} + + case syntax.OpLiteral: + return &equalStringMatcher{ + s: string(re.Rune), + caseSensitive: !isCaseInsensitive(re), + } + case syntax.OpAlternate: + or := make([]StringMatcher, 0, len(re.Sub)) + for _, sub := range re.Sub { + m := stringMatcherFromRegexpInternal(sub) + if m == nil { + return nil + } + or = append(or, m) + } + return orStringMatcher(or) + case syntax.OpConcat: + clearCapture(re.Sub...) + + if len(re.Sub) == 0 { + return emptyStringMatcher{} + } + if len(re.Sub) == 1 { + return stringMatcherFromRegexpInternal(re.Sub[0]) + } + + var left, right StringMatcher + + // Let's try to find if there's a first and last any matchers. + if re.Sub[0].Op == syntax.OpPlus || re.Sub[0].Op == syntax.OpStar || re.Sub[0].Op == syntax.OpQuest { + left = stringMatcherFromRegexpInternal(re.Sub[0]) + if left == nil { + return nil + } + re.Sub = re.Sub[1:] + } + if re.Sub[len(re.Sub)-1].Op == syntax.OpPlus || re.Sub[len(re.Sub)-1].Op == syntax.OpStar || re.Sub[len(re.Sub)-1].Op == syntax.OpQuest { + right = stringMatcherFromRegexpInternal(re.Sub[len(re.Sub)-1]) + if right == nil { + return nil + } + re.Sub = re.Sub[:len(re.Sub)-1] + } + + matches, matchesCaseSensitive := findSetMatchesInternal(re, "") + + if len(matches) == 0 && len(re.Sub) == 2 { + // We have not find fixed set matches. We look for other known cases that + // we can optimize. + switch { + // Prefix is literal. + case right == nil && re.Sub[0].Op == syntax.OpLiteral: + right = stringMatcherFromRegexpInternal(re.Sub[1]) + if right != nil { + matches = []string{string(re.Sub[0].Rune)} + matchesCaseSensitive = !isCaseInsensitive(re.Sub[0]) + } + + // Suffix is literal. + case left == nil && re.Sub[1].Op == syntax.OpLiteral: + left = stringMatcherFromRegexpInternal(re.Sub[0]) + if left != nil { + matches = []string{string(re.Sub[1].Rune)} + matchesCaseSensitive = !isCaseInsensitive(re.Sub[1]) + } + } + } + + // Ensure we've found some literals to match (optionally with a left and/or right matcher). + // If not, then this optimization doesn't trigger. + if len(matches) == 0 { + return nil + } + + // Use the right (and best) matcher based on what we've found. + switch { + // No left and right matchers (only fixed set matches). + case left == nil && right == nil: + // if there's no any matchers on both side it's a concat of literals + or := make([]StringMatcher, 0, len(matches)) + for _, match := range matches { + or = append(or, &equalStringMatcher{ + s: match, + caseSensitive: matchesCaseSensitive, + }) + } + return orStringMatcher(or) + + // Right matcher with 1 fixed set match. + case left == nil && len(matches) == 1: + return &literalPrefixStringMatcher{ + prefix: matches[0], + prefixCaseSensitive: matchesCaseSensitive, + right: right, + } + + // Left matcher with 1 fixed set match. + case right == nil && len(matches) == 1: + return &literalSuffixStringMatcher{ + left: left, + suffix: matches[0], + suffixCaseSensitive: matchesCaseSensitive, + } + + // We found literals in the middle. We can trigger the fast path only if + // the matches are case sensitive because containsStringMatcher doesn't + // support case insensitive. + case matchesCaseSensitive: + return &containsStringMatcher{ + substrings: matches, + left: left, + right: right, + } + } + } + return nil +} + +// containsStringMatcher matches a string if it contains any of the substrings. +// If left and right are not nil, it's a contains operation where left and right must match. +// If left is nil, it's a hasPrefix operation and right must match. +// Finally, if right is nil it's a hasSuffix operation and left must match. +type containsStringMatcher struct { + // The matcher that must match the left side. Can be nil. + left StringMatcher + + // At least one of these strings must match in the "middle", between left and right matchers. + substrings []string + + // The matcher that must match the right side. Can be nil. + right StringMatcher +} + +func (m *containsStringMatcher) Matches(s string) bool { + for _, substr := range m.substrings { + switch { + case m.right != nil && m.left != nil: + searchStartPos := 0 + + for { + pos := strings.Index(s[searchStartPos:], substr) + if pos < 0 { + break + } + + // Since we started searching from searchStartPos, we have to add that offset + // to get the actual position of the substring inside the text. + pos += searchStartPos + + // If both the left and right matchers match, then we can stop searching because + // we've found a match. + if m.left.Matches(s[:pos]) && m.right.Matches(s[pos+len(substr):]) { + return true + } + + // Continue searching for another occurrence of the substring inside the text. + searchStartPos = pos + 1 + } + case m.left != nil: + // If we have to check for characters on the left then we need to match a suffix. + if strings.HasSuffix(s, substr) && m.left.Matches(s[:len(s)-len(substr)]) { + return true + } + case m.right != nil: + if strings.HasPrefix(s, substr) && m.right.Matches(s[len(substr):]) { + return true + } + } + } + return false +} + +// literalPrefixStringMatcher matches a string with the given literal prefix and right side matcher. +type literalPrefixStringMatcher struct { + prefix string + prefixCaseSensitive bool + + // The matcher that must match the right side. Can be nil. + right StringMatcher +} + +func (m *literalPrefixStringMatcher) Matches(s string) bool { + // Ensure the prefix matches. + if m.prefixCaseSensitive && !strings.HasPrefix(s, m.prefix) { + return false + } + if !m.prefixCaseSensitive && !hasPrefixCaseInsensitive(s, m.prefix) { + return false + } + + // Ensure the right side matches. + return m.right.Matches(s[len(m.prefix):]) +} + +// literalSuffixStringMatcher matches a string with the given literal suffix and left side matcher. +type literalSuffixStringMatcher struct { + // The matcher that must match the left side. Can be nil. + left StringMatcher + + suffix string + suffixCaseSensitive bool +} + +func (m *literalSuffixStringMatcher) Matches(s string) bool { + // Ensure the suffix matches. + if m.suffixCaseSensitive && !strings.HasSuffix(s, m.suffix) { + return false + } + if !m.suffixCaseSensitive && !hasSuffixCaseInsensitive(s, m.suffix) { + return false + } + + // Ensure the left side matches. + return m.left.Matches(s[:len(s)-len(m.suffix)]) +} + +// emptyStringMatcher matches an empty string. +type emptyStringMatcher struct{} + +func (m emptyStringMatcher) Matches(s string) bool { + return len(s) == 0 +} + +// orStringMatcher matches any of the sub-matchers. +type orStringMatcher []StringMatcher + +func (m orStringMatcher) Matches(s string) bool { + for _, matcher := range m { + if matcher.Matches(s) { + return true + } + } + return false +} + +// equalStringMatcher matches a string exactly and support case insensitive. +type equalStringMatcher struct { + s string + caseSensitive bool +} + +func (m *equalStringMatcher) Matches(s string) bool { + if m.caseSensitive { + return m.s == s + } + return strings.EqualFold(m.s, s) +} + +type multiStringMatcherBuilder interface { + StringMatcher + add(s string) + setMatches() []string +} + +func newEqualMultiStringMatcher(caseSensitive bool, estimatedSize int) multiStringMatcherBuilder { + // If the estimated size is low enough, it's faster to use a slice instead of a map. + if estimatedSize < minEqualMultiStringMatcherMapThreshold { + return &equalMultiStringSliceMatcher{caseSensitive: caseSensitive, values: make([]string, 0, estimatedSize)} + } + + return &equalMultiStringMapMatcher{ + values: make(map[string]struct{}, estimatedSize), + caseSensitive: caseSensitive, + } +} + +// equalMultiStringSliceMatcher matches a string exactly against a slice of valid values. +type equalMultiStringSliceMatcher struct { + values []string + + caseSensitive bool +} + +func (m *equalMultiStringSliceMatcher) add(s string) { + m.values = append(m.values, s) +} + +func (m *equalMultiStringSliceMatcher) setMatches() []string { + return m.values +} + +func (m *equalMultiStringSliceMatcher) Matches(s string) bool { + if m.caseSensitive { + for _, v := range m.values { + if s == v { + return true + } + } + } else { + for _, v := range m.values { + if strings.EqualFold(s, v) { + return true + } + } + } + return false +} + +// equalMultiStringMapMatcher matches a string exactly against a map of valid values. +type equalMultiStringMapMatcher struct { + // values contains values to match a string against. If the matching is case insensitive, + // the values here must be lowercase. + values map[string]struct{} + + caseSensitive bool +} + +func (m *equalMultiStringMapMatcher) add(s string) { + if !m.caseSensitive { + s = strings.ToLower(s) + } + + m.values[s] = struct{}{} +} + +func (m *equalMultiStringMapMatcher) setMatches() []string { + if len(m.values) >= maxSetMatches { + return nil + } + + matches := make([]string, 0, len(m.values)) + for s := range m.values { + matches = append(matches, s) + } + return matches +} + +func (m *equalMultiStringMapMatcher) Matches(s string) bool { + if !m.caseSensitive { + s = strings.ToLower(s) + } + + _, ok := m.values[s] + return ok +} + +// anyStringWithoutNewlineMatcher is a stringMatcher which matches any string +// (including an empty one) as far as it doesn't contain any newline character. +type anyStringWithoutNewlineMatcher struct{} + +func (m anyStringWithoutNewlineMatcher) Matches(s string) bool { + // We need to make sure it doesn't contain a newline. Since the newline is + // an ASCII character, we can use strings.IndexByte(). + return strings.IndexByte(s, '\n') == -1 +} + +// anyNonEmptyStringMatcher is a stringMatcher which matches any non-empty string. +type anyNonEmptyStringMatcher struct { + matchNL bool +} + +func (m *anyNonEmptyStringMatcher) Matches(s string) bool { + if m.matchNL { + // It's OK if the string contains a newline so we just need to make + // sure it's non-empty. + return len(s) > 0 + } + + // We need to make sure it non-empty and doesn't contain a newline. + // Since the newline is an ASCII character, we can use strings.IndexByte(). + return len(s) > 0 && strings.IndexByte(s, '\n') == -1 +} + +// zeroOrOneCharacterStringMatcher is a StringMatcher which matches zero or one occurrence +// of any character. The newline character is matches only if matchNL is set to true. +type zeroOrOneCharacterStringMatcher struct { + matchNL bool +} + +func (m *zeroOrOneCharacterStringMatcher) Matches(s string) bool { + // Zero or one. + if len(s) > 1 { + return false + } + + // No need to check for the newline if the string is empty or matching a newline is OK. + if m.matchNL || len(s) == 0 { + return true + } + + return s[0] != '\n' +} + +// trueMatcher is a stringMatcher which matches any string (always returns true). +type trueMatcher struct{} + +func (m trueMatcher) Matches(_ string) bool { + return true +} + +// optimizeEqualStringMatchers optimize a specific case where all matchers are made by an +// alternation (orStringMatcher) of strings checked for equality (equalStringMatcher). In +// this specific case, when we have many strings to match against we can use a map instead +// of iterating over the list of strings. +func optimizeEqualStringMatchers(input StringMatcher, threshold int) StringMatcher { + var ( + caseSensitive bool + caseSensitiveSet bool + numValues int + ) + + // Analyse the input StringMatcher to count the number of occurrences + // and ensure all of them have the same case sensitivity. + analyseCallback := func(matcher *equalStringMatcher) bool { + // Ensure we don't have mixed case sensitivity. + if caseSensitiveSet && caseSensitive != matcher.caseSensitive { + return false + } else if !caseSensitiveSet { + caseSensitive = matcher.caseSensitive + caseSensitiveSet = true + } + + numValues++ + return true + } + + if !findEqualStringMatchers(input, analyseCallback) { + return input + } + + // If the number of values found is less than the threshold, then we should skip the optimization. + if numValues < threshold { + return input + } + + // Parse again the input StringMatcher to extract all values and storing them. + // We can skip the case sensitivity check because we've already checked it and + // if the code reach this point then it means all matchers have the same case sensitivity. + multiMatcher := newEqualMultiStringMatcher(caseSensitive, numValues) + + // Ignore the return value because we already iterated over the input StringMatcher + // and it was all good. + findEqualStringMatchers(input, func(matcher *equalStringMatcher) bool { + multiMatcher.add(matcher.s) + return true + }) + + return multiMatcher +} + +// findEqualStringMatchers analyze the input StringMatcher and calls the callback for each +// equalStringMatcher found. Returns true if and only if the input StringMatcher is *only* +// composed by an alternation of equalStringMatcher. +func findEqualStringMatchers(input StringMatcher, callback func(matcher *equalStringMatcher) bool) bool { + orInput, ok := input.(orStringMatcher) + if !ok { + return false + } + + for _, m := range orInput { + switch casted := m.(type) { + case orStringMatcher: + if !findEqualStringMatchers(m, callback) { + return false + } + + case *equalStringMatcher: + if !callback(casted) { + return false + } + + default: + // It's not an equal string matcher, so we have to stop searching + // cause this optimization can't be applied. + return false + } + } + + return true +} + +func hasPrefixCaseInsensitive(s, prefix string) bool { + return len(s) >= len(prefix) && strings.EqualFold(s[0:len(prefix)], prefix) +} + +func hasSuffixCaseInsensitive(s, suffix string) bool { + return len(s) >= len(suffix) && strings.EqualFold(s[len(s)-len(suffix):], suffix) +} diff --git a/model/labels/regexp_test.go b/model/labels/regexp_test.go index 3188f7cefc..3a15b52b40 100644 --- a/model/labels/regexp_test.go +++ b/model/labels/regexp_test.go @@ -14,50 +14,109 @@ package labels import ( + "fmt" + "math/rand" "strings" "testing" + "time" + "github.com/grafana/regexp" "github.com/grafana/regexp/syntax" "github.com/stretchr/testify/require" ) -func TestNewFastRegexMatcher(t *testing.T) { - cases := []struct { - regex string - value string - expected bool - }{ - {regex: "(foo|bar)", value: "foo", expected: true}, - {regex: "(foo|bar)", value: "foo bar", expected: false}, - {regex: "(foo|bar)", value: "bar", expected: true}, - {regex: "foo.*", value: "foo bar", expected: true}, - {regex: "foo.*", value: "bar foo", expected: false}, - {regex: ".*foo", value: "foo bar", expected: false}, - {regex: ".*foo", value: "bar foo", expected: true}, - {regex: ".*foo", value: "foo", expected: true}, - {regex: "^.*foo$", value: "foo", expected: true}, - {regex: "^.+foo$", value: "foo", expected: false}, - {regex: "^.+foo$", value: "bfoo", expected: true}, - {regex: ".*", value: "\n", expected: false}, - {regex: ".*", value: "\nfoo", expected: false}, - {regex: ".*foo", value: "\nfoo", expected: false}, - {regex: "foo.*", value: "foo\n", expected: false}, - {regex: "foo\n.*", value: "foo\n", expected: true}, - {regex: ".*foo.*", value: "foo", expected: true}, - {regex: ".*foo.*", value: "foo bar", expected: true}, - {regex: ".*foo.*", value: "hello foo world", expected: true}, - {regex: ".*foo.*", value: "hello foo\n world", expected: false}, - {regex: ".*foo\n.*", value: "hello foo\n world", expected: true}, - {regex: ".*", value: "foo", expected: true}, - {regex: "", value: "foo", expected: false}, - {regex: "", value: "", expected: true}, +var ( + asciiRunes = []rune("abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789-_") + regexes = []string{ + "", + "foo", + "^foo", + "(foo|bar)", + "foo.*", + ".*foo", + "^.*foo$", + "^.+foo$", + ".*", + ".+", + "foo.+", + ".+foo", + "foo\n.+", + "foo\n.*", + ".*foo.*", + ".+foo.+", + "(?s:.*)", + "(?s:.+)", + "(?s:^.*foo$)", + "(?i:foo)", + "(?i:(foo|bar))", + "(?i:(foo1|foo2|bar))", + "^(?i:foo|oo)|(bar)$", + "(?i:(foo1|foo2|aaa|bbb|ccc|ddd|eee|fff|ggg|hhh|iii|lll|mmm|nnn|ooo|ppp|qqq|rrr|sss|ttt|uuu|vvv|www|xxx|yyy|zzz))", + "((.*)(bar|b|buzz)(.+)|foo)$", + "^$", + "(prometheus|api_prom)_api_v1_.+", + "10\\.0\\.(1|2)\\.+", + "10\\.0\\.(1|2).+", + "((fo(bar))|.+foo)", + // A long case sensitive alternation. + "zQPbMkNO|NNSPdvMi|iWuuSoAl|qbvKMimS|IecrXtPa|seTckYqt|NxnyHkgB|fIDlOgKb|UhlWIygH|OtNoJxHG|cUTkFVIV|mTgFIHjr|jQkoIDtE|PPMKxRXl|AwMfwVkQ|CQyMrTQJ|BzrqxVSi|nTpcWuhF|PertdywG|ZZDgCtXN|WWdDPyyE|uVtNQsKk|BdeCHvPZ|wshRnFlH|aOUIitIp|RxZeCdXT|CFZMslCj|AVBZRDxl|IzIGCnhw|ythYuWiz|oztXVXhl|VbLkwqQx|qvaUgyVC|VawUjPWC|ecloYJuj|boCLTdSU|uPrKeAZx|hrMWLWBq|JOnUNHRM|rYnujkPq|dDEdZhIj|DRrfvugG|yEGfDxVV|YMYdJWuP|PHUQZNWM|AmKNrLis|zTxndVfn|FPsHoJnc|EIulZTua|KlAPhdzg|ScHJJCLt|NtTfMzME|eMCwuFdo|SEpJVJbR|cdhXZeCx|sAVtBwRh|kVFEVcMI|jzJrxraA|tGLHTell|NNWoeSaw|DcOKSetX|UXZAJyka|THpMphDP|rizheevl|kDCBRidd|pCZZRqyu|pSygkitl|SwZGkAaW|wILOrfNX|QkwVOerj|kHOMxPDr|EwOVycJv|AJvtzQFS|yEOjKYYB|LizIINLL|JBRSsfcG|YPiUqqNl|IsdEbvee|MjEpGcBm|OxXZVgEQ|xClXGuxa|UzRCGFEb|buJbvfvA|IPZQxRet|oFYShsMc|oBHffuHO|bzzKrcBR|KAjzrGCl|IPUsAVls|OGMUMbIU|gyDccHuR|bjlalnDd|ZLWjeMna|fdsuIlxQ|dVXtiomV|XxedTjNg|XWMHlNoA|nnyqArQX|opfkWGhb|wYtnhdYb", + // An extremely long case sensitive alternation. This is a special + // case because the values share common prefixes rather than being + // entirely random. This is common in the real world. For example, the + // values of a label like kubernetes pod will often include the + // deployment name as a prefix. + "jyyfj00j0061|jyyfj00j0062|jyyfj94j0093|jyyfj99j0093|jyyfm01j0021|jyyfm02j0021|jyefj00j0192|jyefj00j0193|jyefj00j0194|jyefj00j0195|jyefj00j0196|jyefj00j0197|jyefj00j0290|jyefj00j0291|jyefj00j0292|jyefj00j0293|jyefj00j0294|jyefj00j0295|jyefj00j0296|jyefj00j0297|jyefj89j0394|jyefj90j0394|jyefj91j0394|jyefj95j0347|jyefj96j0322|jyefj96j0347|jyefj97j0322|jyefj97j0347|jyefj98j0322|jyefj98j0347|jyefj99j0320|jyefj99j0322|jyefj99j0323|jyefj99j0335|jyefj99j0336|jyefj99j0344|jyefj99j0347|jyefj99j0349|jyefj99j0351|jyeff00j0117|lyyfm01j0025|lyyfm01j0028|lyyfm01j0041|lyyfm01j0133|lyyfm01j0701|lyyfm02j0025|lyyfm02j0028|lyyfm02j0041|lyyfm02j0133|lyyfm02j0701|lyyfm03j0701|lyefj00j0775|lyefj00j0776|lyefj00j0777|lyefj00j0778|lyefj00j0779|lyefj00j0780|lyefj00j0781|lyefj00j0782|lyefj50j3807|lyefj50j3852|lyefj51j3807|lyefj51j3852|lyefj52j3807|lyefj52j3852|lyefj53j3807|lyefj53j3852|lyefj54j3807|lyefj54j3852|lyefj54j3886|lyefj55j3807|lyefj55j3852|lyefj55j3886|lyefj56j3807|lyefj56j3852|lyefj56j3886|lyefj57j3807|lyefj57j3852|lyefj57j3886|lyefj58j3807|lyefj58j3852|lyefj58j3886|lyefj59j3807|lyefj59j3852|lyefj59j3886|lyefj60j3807|lyefj60j3852|lyefj60j3886|lyefj61j3807|lyefj61j3852|lyefj61j3886|lyefj62j3807|lyefj62j3852|lyefj62j3886|lyefj63j3807|lyefj63j3852|lyefj63j3886|lyefj64j3807|lyefj64j3852|lyefj64j3886|lyefj65j3807|lyefj65j3852|lyefj65j3886|lyefj66j3807|lyefj66j3852|lyefj66j3886|lyefj67j3807|lyefj67j3852|lyefj67j3886|lyefj68j3807|lyefj68j3852|lyefj68j3886|lyefj69j3807|lyefj69j3846|lyefj69j3852|lyefj69j3886|lyefj70j3807|lyefj70j3846|lyefj70j3852|lyefj70j3886|lyefj71j3807|lyefj71j3846|lyefj71j3852|lyefj71j3886|lyefj72j3807|lyefj72j3846|lyefj72j3852|lyefj72j3886|lyefj73j3807|lyefj73j3846|lyefj73j3852|lyefj73j3886|lyefj74j3807|lyefj74j3846|lyefj74j3852|lyefj74j3886|lyefj75j3807|lyefj75j3808|lyefj75j3846|lyefj75j3852|lyefj75j3886|lyefj76j3732|lyefj76j3807|lyefj76j3808|lyefj76j3846|lyefj76j3852|lyefj76j3886|lyefj77j3732|lyefj77j3807|lyefj77j3808|lyefj77j3846|lyefj77j3852|lyefj77j3886|lyefj78j3278|lyefj78j3732|lyefj78j3807|lyefj78j3808|lyefj78j3846|lyefj78j3852|lyefj78j3886|lyefj79j3732|lyefj79j3807|lyefj79j3808|lyefj79j3846|lyefj79j3852|lyefj79j3886|lyefj80j3732|lyefj80j3807|lyefj80j3808|lyefj80j3846|lyefj80j3852|lyefj80j3886|lyefj81j3732|lyefj81j3807|lyefj81j3808|lyefj81j3846|lyefj81j3852|lyefj81j3886|lyefj82j3732|lyefj82j3807|lyefj82j3808|lyefj82j3846|lyefj82j3852|lyefj82j3886|lyefj83j3732|lyefj83j3807|lyefj83j3808|lyefj83j3846|lyefj83j3852|lyefj83j3886|lyefj84j3732|lyefj84j3807|lyefj84j3808|lyefj84j3846|lyefj84j3852|lyefj84j3886|lyefj85j3732|lyefj85j3807|lyefj85j3808|lyefj85j3846|lyefj85j3852|lyefj85j3886|lyefj86j3278|lyefj86j3732|lyefj86j3807|lyefj86j3808|lyefj86j3846|lyefj86j3852|lyefj86j3886|lyefj87j3278|lyefj87j3732|lyefj87j3807|lyefj87j3808|lyefj87j3846|lyefj87j3852|lyefj87j3886|lyefj88j3732|lyefj88j3807|lyefj88j3808|lyefj88j3846|lyefj88j3852|lyefj88j3886|lyefj89j3732|lyefj89j3807|lyefj89j3808|lyefj89j3846|lyefj89j3852|lyefj89j3886|lyefj90j3732|lyefj90j3807|lyefj90j3808|lyefj90j3846|lyefj90j3852|lyefj90j3886|lyefj91j3732|lyefj91j3807|lyefj91j3808|lyefj91j3846|lyefj91j3852|lyefj91j3886|lyefj92j3732|lyefj92j3807|lyefj92j3808|lyefj92j3846|lyefj92j3852|lyefj92j3886|lyefj93j3732|lyefj93j3807|lyefj93j3808|lyefj93j3846|lyefj93j3852|lyefj93j3885|lyefj93j3886|lyefj94j3525|lyefj94j3732|lyefj94j3807|lyefj94j3808|lyefj94j3846|lyefj94j3852|lyefj94j3885|lyefj94j3886|lyefj95j3525|lyefj95j3732|lyefj95j3807|lyefj95j3808|lyefj95j3846|lyefj95j3852|lyefj95j3886|lyefj96j3732|lyefj96j3803|lyefj96j3807|lyefj96j3808|lyefj96j3846|lyefj96j3852|lyefj96j3886|lyefj97j3333|lyefj97j3732|lyefj97j3792|lyefj97j3803|lyefj97j3807|lyefj97j3808|lyefj97j3838|lyefj97j3843|lyefj97j3846|lyefj97j3852|lyefj97j3886|lyefj98j3083|lyefj98j3333|lyefj98j3732|lyefj98j3807|lyefj98j3808|lyefj98j3838|lyefj98j3843|lyefj98j3846|lyefj98j3852|lyefj98j3873|lyefj98j3877|lyefj98j3882|lyefj98j3886|lyefj99j2984|lyefj99j3083|lyefj99j3333|lyefj99j3732|lyefj99j3807|lyefj99j3808|lyefj99j3846|lyefj99j3849|lyefj99j3852|lyefj99j3873|lyefj99j3877|lyefj99j3882|lyefj99j3884|lyefj99j3886|lyeff00j0106|lyeff00j0107|lyeff00j0108|lyeff00j0129|lyeff00j0130|lyeff00j0131|lyeff00j0132|lyeff00j0133|lyeff00j0134|lyeff00j0444|lyeff00j0445|lyeff91j0473|lyeff92j0473|lyeff92j3877|lyeff93j3877|lyeff94j0501|lyeff94j3525|lyeff94j3877|lyeff95j0501|lyeff95j3525|lyeff95j3877|lyeff96j0503|lyeff96j3877|lyeff97j3877|lyeff98j3333|lyeff98j3877|lyeff99j2984|lyeff99j3333|lyeff99j3877|mfyr9149ej|mfyr9149ek|mfyr9156ej|mfyr9156ek|mfyr9157ej|mfyr9157ek|mfyr9159ej|mfyr9159ek|mfyr9203ej|mfyr9204ej|mfyr9205ej|mfyr9206ej|mfyr9207ej|mfyr9207ek|mfyr9217ej|mfyr9217ek|mfyr9222ej|mfyr9222ek|mfyu0185ej|mfye9187ej|mfye9187ek|mfye9188ej|mfye9188ek|mfye9189ej|mfye9189ek|mfyf0185ej|oyefj87j0007|oyefj88j0007|oyefj89j0007|oyefj90j0007|oyefj91j0007|oyefj95j0001|oyefj96j0001|oyefj98j0004|oyefj99j0004|oyeff91j0004|oyeff92j0004|oyeff93j0004|oyeff94j0004|oyeff95j0004|oyeff96j0004|rklvyaxmany|ryefj93j0001|ryefj94j0001|tyyfj00a0001|tyyfj84j0005|tyyfj85j0005|tyyfj86j0005|tyyfj87j0005|tyyfj88j0005|tyyfj89j0005|tyyfj90j0005|tyyfj91j0005|tyyfj92j0005|tyyfj93j0005|tyyfj94j0005|tyyfj95j0005|tyyfj96j0005|tyyfj97j0005|tyyfj98j0005|tyyfj99j0005|tyefj50j0015|tyefj50j0017|tyefj50j0019|tyefj50j0020|tyefj50j0021|tyefj51j0015|tyefj51j0017|tyefj51j0019|tyefj51j0020|tyefj51j0021|tyefj52j0015|tyefj52j0017|tyefj52j0019|tyefj52j0020|tyefj52j0021|tyefj53j0015|tyefj53j0017|tyefj53j0019|tyefj53j0020|tyefj53j0021|tyefj54j0015|tyefj54j0017|tyefj54j0019|tyefj54j0020|tyefj54j0021|tyefj55j0015|tyefj55j0017|tyefj55j0019|tyefj55j0020|tyefj55j0021|tyefj56j0015|tyefj56j0017|tyefj56j0019|tyefj56j0020|tyefj56j0021|tyefj57j0015|tyefj57j0017|tyefj57j0019|tyefj57j0020|tyefj57j0021|tyefj58j0015|tyefj58j0017|tyefj58j0019|tyefj58j0020|tyefj58j0021|tyefj59j0015|tyefj59j0017|tyefj59j0019|tyefj59j0020|tyefj59j0021|tyefj60j0015|tyefj60j0017|tyefj60j0019|tyefj60j0020|tyefj60j0021|tyefj61j0015|tyefj61j0017|tyefj61j0019|tyefj61j0020|tyefj61j0021|tyefj62j0015|tyefj62j0017|tyefj62j0019|tyefj62j0020|tyefj62j0021|tyefj63j0015|tyefj63j0017|tyefj63j0019|tyefj63j0020|tyefj63j0021|tyefj64j0015|tyefj64j0017|tyefj64j0019|tyefj64j0020|tyefj64j0021|tyefj65j0015|tyefj65j0017|tyefj65j0019|tyefj65j0020|tyefj65j0021|tyefj66j0015|tyefj66j0017|tyefj66j0019|tyefj66j0020|tyefj66j0021|tyefj67j0015|tyefj67j0017|tyefj67j0019|tyefj67j0020|tyefj67j0021|tyefj68j0015|tyefj68j0017|tyefj68j0019|tyefj68j0020|tyefj68j0021|tyefj69j0015|tyefj69j0017|tyefj69j0019|tyefj69j0020|tyefj69j0021|tyefj70j0015|tyefj70j0017|tyefj70j0019|tyefj70j0020|tyefj70j0021|tyefj71j0015|tyefj71j0017|tyefj71j0019|tyefj71j0020|tyefj71j0021|tyefj72j0015|tyefj72j0017|tyefj72j0019|tyefj72j0020|tyefj72j0021|tyefj72j0022|tyefj73j0015|tyefj73j0017|tyefj73j0019|tyefj73j0020|tyefj73j0021|tyefj73j0022|tyefj74j0015|tyefj74j0017|tyefj74j0019|tyefj74j0020|tyefj74j0021|tyefj74j0022|tyefj75j0015|tyefj75j0017|tyefj75j0019|tyefj75j0020|tyefj75j0021|tyefj75j0022|tyefj76j0015|tyefj76j0017|tyefj76j0019|tyefj76j0020|tyefj76j0021|tyefj76j0022|tyefj76j0119|tyefj77j0015|tyefj77j0017|tyefj77j0019|tyefj77j0020|tyefj77j0021|tyefj77j0022|tyefj77j0119|tyefj78j0015|tyefj78j0017|tyefj78j0019|tyefj78j0020|tyefj78j0021|tyefj78j0022|tyefj78j0119|tyefj79j0015|tyefj79j0017|tyefj79j0019|tyefj79j0020|tyefj79j0021|tyefj79j0022|tyefj79j0119|tyefj80j0015|tyefj80j0017|tyefj80j0019|tyefj80j0020|tyefj80j0021|tyefj80j0022|tyefj80j0114|tyefj80j0119|tyefj81j0015|tyefj81j0017|tyefj81j0019|tyefj81j0020|tyefj81j0021|tyefj81j0022|tyefj81j0114|tyefj81j0119|tyefj82j0015|tyefj82j0017|tyefj82j0019|tyefj82j0020|tyefj82j0021|tyefj82j0022|tyefj82j0119|tyefj83j0015|tyefj83j0017|tyefj83j0019|tyefj83j0020|tyefj83j0021|tyefj83j0022|tyefj83j0119|tyefj84j0014|tyefj84j0015|tyefj84j0017|tyefj84j0019|tyefj84j0020|tyefj84j0021|tyefj84j0022|tyefj84j0119|tyefj85j0014|tyefj85j0015|tyefj85j0017|tyefj85j0019|tyefj85j0020|tyefj85j0021|tyefj85j0022|tyefj85j0119|tyefj86j0014|tyefj86j0015|tyefj86j0017|tyefj86j0019|tyefj86j0020|tyefj86j0021|tyefj86j0022|tyefj87j0014|tyefj87j0015|tyefj87j0017|tyefj87j0019|tyefj87j0020|tyefj87j0021|tyefj87j0022|tyefj88j0014|tyefj88j0015|tyefj88j0017|tyefj88j0019|tyefj88j0020|tyefj88j0021|tyefj88j0022|tyefj88j0100|tyefj88j0115|tyefj89j0003|tyefj89j0014|tyefj89j0015|tyefj89j0017|tyefj89j0019|tyefj89j0020|tyefj89j0021|tyefj89j0022|tyefj89j0100|tyefj89j0115|tyefj90j0014|tyefj90j0015|tyefj90j0016|tyefj90j0017|tyefj90j0018|tyefj90j0019|tyefj90j0020|tyefj90j0021|tyefj90j0022|tyefj90j0100|tyefj90j0111|tyefj90j0115|tyefj91j0014|tyefj91j0015|tyefj91j0016|tyefj91j0017|tyefj91j0018|tyefj91j0019|tyefj91j0020|tyefj91j0021|tyefj91j0022|tyefj91j0100|tyefj91j0111|tyefj91j0115|tyefj92j0014|tyefj92j0015|tyefj92j0016|tyefj92j0017|tyefj92j0018|tyefj92j0019|tyefj92j0020|tyefj92j0021|tyefj92j0022|tyefj92j0100|tyefj92j0105|tyefj92j0115|tyefj92j0121|tyefj93j0004|tyefj93j0014|tyefj93j0015|tyefj93j0017|tyefj93j0018|tyefj93j0019|tyefj93j0020|tyefj93j0021|tyefj93j0022|tyefj93j0100|tyefj93j0105|tyefj93j0115|tyefj93j0121|tyefj94j0002|tyefj94j0004|tyefj94j0008|tyefj94j0014|tyefj94j0015|tyefj94j0017|tyefj94j0019|tyefj94j0020|tyefj94j0021|tyefj94j0022|tyefj94j0084|tyefj94j0088|tyefj94j0100|tyefj94j0106|tyefj94j0116|tyefj94j0121|tyefj94j0123|tyefj95j0002|tyefj95j0004|tyefj95j0008|tyefj95j0014|tyefj95j0015|tyefj95j0017|tyefj95j0019|tyefj95j0020|tyefj95j0021|tyefj95j0022|tyefj95j0084|tyefj95j0088|tyefj95j0100|tyefj95j0101|tyefj95j0106|tyefj95j0112|tyefj95j0116|tyefj95j0121|tyefj95j0123|tyefj96j0014|tyefj96j0015|tyefj96j0017|tyefj96j0019|tyefj96j0020|tyefj96j0021|tyefj96j0022|tyefj96j0082|tyefj96j0084|tyefj96j0100|tyefj96j0101|tyefj96j0112|tyefj96j0117|tyefj96j0121|tyefj96j0124|tyefj97j0014|tyefj97j0015|tyefj97j0017|tyefj97j0019|tyefj97j0020|tyefj97j0021|tyefj97j0022|tyefj97j0081|tyefj97j0087|tyefj97j0098|tyefj97j0100|tyefj97j0107|tyefj97j0109|tyefj97j0113|tyefj97j0117|tyefj97j0118|tyefj97j0121|tyefj98j0003|tyefj98j0006|tyefj98j0014|tyefj98j0015|tyefj98j0017|tyefj98j0019|tyefj98j0020|tyefj98j0021|tyefj98j0022|tyefj98j0083|tyefj98j0085|tyefj98j0086|tyefj98j0100|tyefj98j0104|tyefj98j0118|tyefj98j0121|tyefj99j0003|tyefj99j0006|tyefj99j0007|tyefj99j0014|tyefj99j0015|tyefj99j0017|tyefj99j0019|tyefj99j0020|tyefj99j0021|tyefj99j0022|tyefj99j0023|tyefj99j0100|tyefj99j0108|tyefj99j0110|tyefj99j0121|tyefj99j0125|tyeff94j0002|tyeff94j0008|tyeff94j0010|tyeff94j0011|tyeff94j0035|tyeff95j0002|tyeff95j0006|tyeff95j0008|tyeff95j0010|tyeff95j0011|tyeff95j0035|tyeff96j0003|tyeff96j0006|tyeff96j0009|tyeff96j0010|tyeff97j0004|tyeff97j0009|tyeff97j0116|tyeff98j0007|tyeff99j0007|tyeff99j0125|uyyfj00j0484|uyyfj00j0485|uyyfj00j0486|uyyfj00j0487|uyyfj00j0488|uyyfj00j0489|uyyfj00j0490|uyyfj00j0491|uyyfj00j0492|uyyfj00j0493|uyyfj00j0494|uyyfj00j0495|uyyfj00j0496|uyyfj00j0497|uyyfj00j0498|uyyfj00j0499|uyyfj00j0500|uyyfj00j0501|uyyfj00j0502|uyyfj00j0503|uyyfj00j0504|uyyfj00j0505|uyyfj00j0506|uyyfj00j0507|uyyfj00j0508|uyyfj00j0509|uyyfj00j0510|uyyfj00j0511|uyyfj00j0512|uyyfj00j0513|uyyfj00j0514|uyyfj00j0515|uyyfj00j0516|uyyfj00j0517|uyyfj00j0518|uyyfj00j0519|uyyfj00j0520|uyyfj00j0521|uyyfj00j0522|uyyfj00j0523|uyyfj00j0524|uyyfj00j0525|uyyfj00j0526|uyyfj00j0527|uyyfj00j0528|uyyfj00j0529|uyyfj00j0530|uyyfj00j0531|uyyfj00j0532|uyyfj00j0533|uyyfj00j0534|uyyfj00j0535|uyyfj00j0536|uyyfj00j0537|uyyfj00j0538|uyyfj00j0539|uyyfj00j0540|uyyfj00j0541|uyyfj00j0542|uyyfj00j0543|uyyfj00j0544|uyyfj00j0545|uyyfj00j0546|uyyfj00j0547|uyyfj00j0548|uyyfj00j0549|uyyfj00j0550|uyyfj00j0551|uyyfj00j0553|uyyfj00j0554|uyyfj00j0555|uyyfj00j0556|uyyfj00j0557|uyyfj00j0558|uyyfj00j0559|uyyfj00j0560|uyyfj00j0561|uyyfj00j0562|uyyfj00j0563|uyyfj00j0564|uyyfj00j0565|uyyfj00j0566|uyyfj00j0614|uyyfj00j0615|uyyfj00j0616|uyyfj00j0617|uyyfj00j0618|uyyfj00j0619|uyyfj00j0620|uyyfj00j0621|uyyfj00j0622|uyyfj00j0623|uyyfj00j0624|uyyfj00j0625|uyyfj00j0626|uyyfj00j0627|uyyfj00j0628|uyyfj00j0629|uyyfj00j0630|uyyfj00j0631|uyyfj00j0632|uyyfj00j0633|uyyfj00j0634|uyyfj00j0635|uyyfj00j0636|uyyfj00j0637|uyyfj00j0638|uyyfj00j0639|uyyfj00j0640|uyyfj00j0641|uyyfj00j0642|uyyfj00j0643|uyyfj00j0644|uyyfj00j0645|uyyfj00j0646|uyyfj00j0647|uyyfj00j0648|uyyfj00j0649|uyyfj00j0650|uyyfj00j0651|uyyfj00j0652|uyyfj00j0653|uyyfj00j0654|uyyfj00j0655|uyyfj00j0656|uyyfj00j0657|uyyfj00j0658|uyyfj00j0659|uyyfj00j0660|uyyfj00j0661|uyyfj00j0662|uyyfj00j0663|uyyfj00j0664|uyyfj00j0665|uyyfj00j0666|uyyfj00j0667|uyyfj00j0668|uyyfj00j0669|uyyfj00j0670|uyyfj00j0671|uyyfj00j0672|uyyfj00j0673|uyyfj00j0674|uyyfj00j0675|uyyfj00j0676|uyyfj00j0677|uyyfj00j0678|uyyfj00j0679|uyyfj00j0680|uyyfj00j0681|uyyfj00j0682|uyyfj00j0683|uyyfj00j0684|uyyfj00j0685|uyyfj00j0686|uyyfj00j0687|uyyfj00j0688|uyyfj00j0689|uyyfj00j0690|uyyfj00j0691|uyyfj00j0692|uyyfj00j0693|uyyfj00j0694|uyyfj00j0695|uyyfj00j0696|uyyfj00j0697|uyyfj00j0698|uyyfj00j0699|uyyfj00j0700|uyyfj00j0701|uyyfj00j0702|uyyfj00j0703|uyyfj00j0704|uyyfj00j0705|uyyfj00j0706|uyyfj00j0707|uyyfj00j0708|uyyfj00j0709|uyyfj00j0710|uyyfj00j0711|uyyfj00j0712|uyyfj00j0713|uyyfj00j0714|uyyfj00j0715|uyyfj00j0716|uyyfj00j0717|uyyfj00j0718|uyyfj00j0719|uyyfj00j0720|uyyfj00j0721|uyyfj00j0722|uyyfj00j0723|uyyfj00j0724|uyyfj00j0725|uyyfj00j0726|uyyfj00j0727|uyyfj00j0728|uyyfj00j0729|uyyfj00j0730|uyyfj00j0731|uyyfj00j0732|uyyfj00j0733|uyyfj00j0734|uyyfj00j0735|uyyfj00j0736|uyyfj00j0737|uyyfj00j0738|uyyfj00j0739|uyyfj00j0740|uyyfj00j0741|uyyfj00j0742|uyyfj00j0743|uyyfj00j0744|uyyfj00j0745|uyyfj00j0746|uyyfj00j0747|uyyfj00j0748|uyyfj00j0749|uyyfj00j0750|uyyfj00j0751|uyyfj00j0752|uyyfj00j0753|uyyfj00j0754|uyyfj00j0755|uyyfj00j0756|uyyfj00j0757|uyyfj00j0758|uyyfj00j0759|uyyfj00j0760|uyyfj00j0761|uyyfj00j0762|uyyfj00j0763|uyyfj00j0764|uyyfj00j0765|uyyfj00j0766|uyyfj00j0767|uyyfj00j0768|uyyfj00j0769|uyyfj00j0770|uyyfj00j0771|uyyfj00j0772|uyyfj00j0773|uyyfj00j0774|uyyfj00j0775|uyyfj00j0776|uyyfj00j0777|uyyfj00j0778|uyyfj00j0779|uyyfj00j0780|uyyfj00j0781|uyyfj00j0782|uyyff00j0011|uyyff00j0031|uyyff00j0032|uyyff00j0033|uyyff00j0034|uyyff99j0012|uyefj00j0071|uyefj00j0455|uyefj00j0456|uyefj00j0582|uyefj00j0583|uyefj00j0584|uyefj00j0585|uyefj00j0586|uyefj00j0590|uyeff00j0188|xyrly-f-jyy-y01|xyrly-f-jyy-y02|xyrly-f-jyy-y03|xyrly-f-jyy-y04|xyrly-f-jyy-y05|xyrly-f-jyy-y06|xyrly-f-jyy-y07|xyrly-f-jyy-y08|xyrly-f-jyy-y09|xyrly-f-jyy-y10|xyrly-f-jyy-y11|xyrly-f-jyy-y12|xyrly-f-jyy-y13|xyrly-f-jyy-y14|xyrly-f-jyy-y15|xyrly-f-jyy-y16|xyrly-f-url-y01|xyrly-f-url-y02|yyefj97j0005|ybyfcy4000|ybyfcy4001|ayefj99j0035|by-b-y-bzu-l01|by-b-y-bzu-l02|by-b-e-079|by-b-e-080|by-b-e-082|by-b-e-083|byefj72j0002|byefj73j0002|byefj74j0002|byefj75j0002|byefj76j0002|byefj77j0002|byefj78j0002|byefj79j0002|byefj91j0007|byefj92j0007|byefj98j0003|byefj99j0003|byefj99j0005|byefj99j0006|byeff88j0002|byeff89j0002|byeff90j0002|byeff91j0002|byeff92j0002|byeff93j0002|byeff96j0003|byeff97j0003|byeff98j0003|byeff99j0003|fymfj98j0001|fymfj99j0001|fyyaj98k0297|fyyaj99k0297|fyyfj00j0109|fyyfj00j0110|fyyfj00j0122|fyyfj00j0123|fyyfj00j0201|fyyfj00j0202|fyyfj00j0207|fyyfj00j0208|fyyfj00j0227|fyyfj00j0228|fyyfj00j0229|fyyfj00j0230|fyyfj00j0231|fyyfj00j0232|fyyfj00j0233|fyyfj00j0234|fyyfj00j0235|fyyfj00j0236|fyyfj00j0237|fyyfj00j0238|fyyfj00j0239|fyyfj00j0240|fyyfj00j0241|fyyfj00j0242|fyyfj00j0243|fyyfj00j0244|fyyfj00j0245|fyyfj00j0246|fyyfj00j0247|fyyfj00j0248|fyyfj00j0249|fyyfj00j0250|fyyfj00j0251|fyyfj00j0252|fyyfj00j0253|fyyfj00j0254|fyyfj00j0255|fyyfj00j0256|fyyfj00j0257|fyyfj00j0258|fyyfj00j0259|fyyfj00j0260|fyyfj00j0261|fyyfj00j0262|fyyfj00j0263|fyyfj00j0264|fyyfj00j0265|fyyfj00j0266|fyyfj00j0267|fyyfj00j0268|fyyfj00j0290|fyyfj00j0291|fyyfj00j0292|fyyfj00j0293|fyyfj00j0294|fyyfj00j0295|fyyfj00j0296|fyyfj00j0297|fyyfj00j0298|fyyfj00j0299|fyyfj00j0300|fyyfj00j0301|fyyfj00j0302|fyyfj00j0303|fyyfj00j0304|fyyfj00j0305|fyyfj00j0306|fyyfj00j0307|fyyfj00j0308|fyyfj00j0309|fyyfj00j0310|fyyfj00j0311|fyyfj00j0312|fyyfj00j0313|fyyfj00j0314|fyyfj00j0315|fyyfj00j0316|fyyfj00j0317|fyyfj00j0318|fyyfj00j0319|fyyfj00j0320|fyyfj00j0321|fyyfj00j0322|fyyfj00j0323|fyyfj00j0324|fyyfj00j0325|fyyfj00j0326|fyyfj00j0327|fyyfj00j0328|fyyfj00j0329|fyyfj00j0330|fyyfj00j0331|fyyfj00j0332|fyyfj00j0333|fyyfj00j0334|fyyfj00j0335|fyyfj00j0340|fyyfj00j0341|fyyfj00j0342|fyyfj00j0343|fyyfj00j0344|fyyfj00j0345|fyyfj00j0346|fyyfj00j0347|fyyfj00j0348|fyyfj00j0349|fyyfj00j0367|fyyfj00j0368|fyyfj00j0369|fyyfj00j0370|fyyfj00j0371|fyyfj00j0372|fyyfj00j0373|fyyfj00j0374|fyyfj00j0375|fyyfj00j0376|fyyfj00j0377|fyyfj00j0378|fyyfj00j0379|fyyfj00j0380|fyyfj00j0381|fyyfj00j0382|fyyfj00j0383|fyyfj00j0384|fyyfj00j0385|fyyfj00j0386|fyyfj00j0387|fyyfj00j0388|fyyfj00j0415|fyyfj00j0416|fyyfj00j0417|fyyfj00j0418|fyyfj00j0419|fyyfj00j0420|fyyfj00j0421|fyyfj00j0422|fyyfj00j0423|fyyfj00j0424|fyyfj00j0425|fyyfj00j0426|fyyfj00j0427|fyyfj00j0428|fyyfj00j0429|fyyfj00j0430|fyyfj00j0431|fyyfj00j0432|fyyfj00j0433|fyyfj00j0434|fyyfj00j0435|fyyfj00j0436|fyyfj00j0437|fyyfj00j0438|fyyfj00j0439|fyyfj00j0440|fyyfj00j0441|fyyfj00j0446|fyyfj00j0447|fyyfj00j0448|fyyfj00j0449|fyyfj00j0451|fyyfj00j0452|fyyfj00j0453|fyyfj00j0454|fyyfj00j0455|fyyfj00j0456|fyyfj00j0457|fyyfj00j0459|fyyfj00j0460|fyyfj00j0461|fyyfj00j0462|fyyfj00j0463|fyyfj00j0464|fyyfj00j0465|fyyfj00j0466|fyyfj00j0467|fyyfj00j0468|fyyfj00j0469|fyyfj00j0470|fyyfj00j0471|fyyfj00j0474|fyyfj00j0475|fyyfj00j0476|fyyfj00j0477|fyyfj00j0478|fyyfj00j0479|fyyfj00j0480|fyyfj00j0481|fyyfj00j0482|fyyfj00j0483|fyyfj00j0484|fyyfj00j0485|fyyfj00j0486|fyyfj00j0487|fyyfj00j0488|fyyfj00j0489|fyyfj00j0490|fyyfj00j0491|fyyfj00j0492|fyyfj00j0493|fyyfj00j0494|fyyfj00j0495|fyyfj00j0496|fyyfj00j0497|fyyfj00j0498|fyyfj00j0499|fyyfj00j0500|fyyfj00j0501|fyyfj00j0502|fyyfj00j0503|fyyfj00j0504|fyyfj00j0505|fyyfj00j0506|fyyfj00j0507|fyyfj00j0508|fyyfj00j0509|fyyfj00j0510|fyyfj00j0511|fyyfj00j0512|fyyfj00j0513|fyyfj00j0514|fyyfj00j0515|fyyfj00j0516|fyyfj00j0517|fyyfj00j0518|fyyfj00j0521|fyyfj00j0522|fyyfj00j0523|fyyfj00j0524|fyyfj00j0526|fyyfj00j0527|fyyfj00j0528|fyyfj00j0529|fyyfj00j0530|fyyfj00j0531|fyyfj00j0532|fyyfj00j0533|fyyfj00j0534|fyyfj00j0535|fyyfj00j0536|fyyfj00j0537|fyyfj00j0538|fyyfj00j0539|fyyfj00j0540|fyyfj00j0541|fyyfj00j0542|fyyfj00j0543|fyyfj00j0544|fyyfj00j0545|fyyfj00j0546|fyyfj00j0564|fyyfj00j0565|fyyfj00j0566|fyyfj00j0567|fyyfj00j0568|fyyfj00j0569|fyyfj00j0570|fyyfj00j0571|fyyfj00j0572|fyyfj00j0574|fyyfj00j0575|fyyfj00j0576|fyyfj00j0577|fyyfj00j0578|fyyfj00j0579|fyyfj00j0580|fyyfj01j0473|fyyfj02j0473|fyyfj36j0289|fyyfj37j0209|fyyfj37j0289|fyyfj38j0209|fyyfj38j0289|fyyfj39j0209|fyyfj39j0289|fyyfj40j0209|fyyfj40j0289|fyyfj41j0209|fyyfj41j0289|fyyfj42j0209|fyyfj42j0289|fyyfj43j0209|fyyfj43j0289|fyyfj44j0209|fyyfj44j0289|fyyfj45j0104|fyyfj45j0209|fyyfj45j0289|fyyfj46j0104|fyyfj46j0209|fyyfj46j0289|fyyfj47j0104|fyyfj47j0209|fyyfj47j0289|fyyfj48j0104|fyyfj48j0209|fyyfj48j0289|fyyfj49j0104|fyyfj49j0209|fyyfj49j0289|fyyfj50j0104|fyyfj50j0209|fyyfj50j0289|fyyfj50j0500|fyyfj51j0104|fyyfj51j0209|fyyfj51j0289|fyyfj51j0500|fyyfj52j0104|fyyfj52j0209|fyyfj52j0289|fyyfj52j0500|fyyfj53j0104|fyyfj53j0209|fyyfj53j0289|fyyfj53j0500|fyyfj54j0104|fyyfj54j0209|fyyfj54j0289|fyyfj54j0500|fyyfj55j0104|fyyfj55j0209|fyyfj55j0289|fyyfj55j0500|fyyfj56j0104|fyyfj56j0209|fyyfj56j0289|fyyfj56j0500|fyyfj57j0104|fyyfj57j0209|fyyfj57j0289|fyyfj57j0500|fyyfj58j0104|fyyfj58j0209|fyyfj58j0289|fyyfj58j0500|fyyfj59j0104|fyyfj59j0209|fyyfj59j0289|fyyfj59j0500|fyyfj60j0104|fyyfj60j0209|fyyfj60j0289|fyyfj60j0500|fyyfj61j0104|fyyfj61j0209|fyyfj61j0289|fyyfj61j0500|fyyfj62j0104|fyyfj62j0209|fyyfj62j0289|fyyfj62j0500|fyyfj63j0104|fyyfj63j0209|fyyfj63j0289|fyyfj63j0500|fyyfj64j0104|fyyfj64j0107|fyyfj64j0209|fyyfj64j0289|fyyfj64j0500|fyyfj64j0573|fyyfj65j0104|fyyfj65j0107|fyyfj65j0209|fyyfj65j0289|fyyfj65j0500|fyyfj65j0573|fyyfj66j0104|fyyfj66j0107|fyyfj66j0209|fyyfj66j0289|fyyfj66j0500|fyyfj66j0573|fyyfj67j0104|fyyfj67j0107|fyyfj67j0209|fyyfj67j0289|fyyfj67j0500|fyyfj67j0573|fyyfj68j0104|fyyfj68j0107|fyyfj68j0209|fyyfj68j0289|fyyfj68j0500|fyyfj68j0573|fyyfj69j0104|fyyfj69j0107|fyyfj69j0209|fyyfj69j0289|fyyfj69j0500|fyyfj69j0573|fyyfj70j0104|fyyfj70j0107|fyyfj70j0209|fyyfj70j0289|fyyfj70j0472|fyyfj70j0500|fyyfj70j0573|fyyfj71j0104|fyyfj71j0107|fyyfj71j0209|fyyfj71j0289|fyyfj71j0472|fyyfj71j0500|fyyfj71j0573|fyyfj72j0104|fyyfj72j0107|fyyfj72j0209|fyyfj72j0289|fyyfj72j0472|fyyfj72j0500|fyyfj72j0573|fyyfj73j0104|fyyfj73j0107|fyyfj73j0209|fyyfj73j0289|fyyfj73j0472|fyyfj73j0500|fyyfj73j0573|fyyfj74j0104|fyyfj74j0107|fyyfj74j0209|fyyfj74j0289|fyyfj74j0472|fyyfj74j0500|fyyfj74j0573|fyyfj75j0104|fyyfj75j0107|fyyfj75j0108|fyyfj75j0209|fyyfj75j0289|fyyfj75j0472|fyyfj75j0500|fyyfj75j0573|fyyfj76j0104|fyyfj76j0107|fyyfj76j0108|fyyfj76j0209|fyyfj76j0289|fyyfj76j0472|fyyfj76j0500|fyyfj76j0573|fyyfj77j0104|fyyfj77j0107|fyyfj77j0108|fyyfj77j0209|fyyfj77j0289|fyyfj77j0472|fyyfj77j0500|fyyfj77j0573|fyyfj78j0104|fyyfj78j0107|fyyfj78j0108|fyyfj78j0209|fyyfj78j0289|fyyfj78j0472|fyyfj78j0500|fyyfj78j0573|fyyfj79j0104|fyyfj79j0107|fyyfj79j0108|fyyfj79j0209|fyyfj79j0289|fyyfj79j0339|fyyfj79j0472|fyyfj79j0500|fyyfj79j0573|fyyfj80j0104|fyyfj80j0107|fyyfj80j0108|fyyfj80j0209|fyyfj80j0289|fyyfj80j0339|fyyfj80j0352|fyyfj80j0472|fyyfj80j0500|fyyfj80j0573|fyyfj81j0104|fyyfj81j0107|fyyfj81j0108|fyyfj81j0209|fyyfj81j0289|fyyfj81j0339|fyyfj81j0352|fyyfj81j0472|fyyfj81j0500|fyyfj81j0573|fyyfj82j0104|fyyfj82j0107|fyyfj82j0108|fyyfj82j0209|fyyfj82j0289|fyyfj82j0339|fyyfj82j0352|fyyfj82j0472|fyyfj82j0500|fyyfj82j0573|fyyfj83j0104|fyyfj83j0107|fyyfj83j0108|fyyfj83j0209|fyyfj83j0289|fyyfj83j0339|fyyfj83j0352|fyyfj83j0472|fyyfj83j0500|fyyfj83j0573|fyyfj84j0104|fyyfj84j0107|fyyfj84j0108|fyyfj84j0209|fyyfj84j0289|fyyfj84j0339|fyyfj84j0352|fyyfj84j0472|fyyfj84j0500|fyyfj84j0573|fyyfj85j0104|fyyfj85j0107|fyyfj85j0108|fyyfj85j0209|fyyfj85j0289|fyyfj85j0301|fyyfj85j0339|fyyfj85j0352|fyyfj85j0472|fyyfj85j0500|fyyfj85j0573|fyyfj86j0104|fyyfj86j0107|fyyfj86j0108|fyyfj86j0209|fyyfj86j0289|fyyfj86j0301|fyyfj86j0339|fyyfj86j0352|fyyfj86j0472|fyyfj86j0500|fyyfj86j0573|fyyfj87j0067|fyyfj87j0104|fyyfj87j0107|fyyfj87j0108|fyyfj87j0209|fyyfj87j0289|fyyfj87j0301|fyyfj87j0339|fyyfj87j0352|fyyfj87j0472|fyyfj87j0500|fyyfj87j0573|fyyfj88j0067|fyyfj88j0104|fyyfj88j0107|fyyfj88j0108|fyyfj88j0209|fyyfj88j0289|fyyfj88j0301|fyyfj88j0339|fyyfj88j0352|fyyfj88j0472|fyyfj88j0500|fyyfj88j0573|fyyfj89j0067|fyyfj89j0104|fyyfj89j0107|fyyfj89j0108|fyyfj89j0209|fyyfj89j0289|fyyfj89j0301|fyyfj89j0339|fyyfj89j0352|fyyfj89j0358|fyyfj89j0472|fyyfj89j0500|fyyfj89j0573|fyyfj90j0067|fyyfj90j0104|fyyfj90j0107|fyyfj90j0108|fyyfj90j0209|fyyfj90j0289|fyyfj90j0301|fyyfj90j0321|fyyfj90j0339|fyyfj90j0352|fyyfj90j0358|fyyfj90j0452|fyyfj90j0472|fyyfj90j0500|fyyfj90j0573|fyyfj91j0067|fyyfj91j0104|fyyfj91j0107|fyyfj91j0108|fyyfj91j0209|fyyfj91j0289|fyyfj91j0301|fyyfj91j0321|fyyfj91j0339|fyyfj91j0352|fyyfj91j0358|fyyfj91j0452|fyyfj91j0472|fyyfj91j0500|fyyfj91j0573|fyyfj92j0067|fyyfj92j0104|fyyfj92j0107|fyyfj92j0108|fyyfj92j0209|fyyfj92j0289|fyyfj92j0301|fyyfj92j0321|fyyfj92j0339|fyyfj92j0352|fyyfj92j0358|fyyfj92j0452|fyyfj92j0472|fyyfj92j0500|fyyfj92j0573|fyyfj93j0067|fyyfj93j0099|fyyfj93j0104|fyyfj93j0107|fyyfj93j0108|fyyfj93j0209|fyyfj93j0289|fyyfj93j0301|fyyfj93j0321|fyyfj93j0352|fyyfj93j0358|fyyfj93j0452|fyyfj93j0472|fyyfj93j0500|fyyfj93j0573|fyyfj94j0067|fyyfj94j0099|fyyfj94j0104|fyyfj94j0107|fyyfj94j0108|fyyfj94j0209|fyyfj94j0211|fyyfj94j0289|fyyfj94j0301|fyyfj94j0321|fyyfj94j0352|fyyfj94j0358|fyyfj94j0359|fyyfj94j0452|fyyfj94j0472|fyyfj94j0500|fyyfj94j0573|fyyfj95j0067|fyyfj95j0099|fyyfj95j0104|fyyfj95j0107|fyyfj95j0108|fyyfj95j0209|fyyfj95j0211|fyyfj95j0289|fyyfj95j0298|fyyfj95j0301|fyyfj95j0321|fyyfj95j0339|fyyfj95j0352|fyyfj95j0358|fyyfj95j0359|fyyfj95j0414|fyyfj95j0452|fyyfj95j0472|fyyfj95j0500|fyyfj95j0573|fyyfj96j0067|fyyfj96j0099|fyyfj96j0104|fyyfj96j0107|fyyfj96j0108|fyyfj96j0209|fyyfj96j0211|fyyfj96j0289|fyyfj96j0298|fyyfj96j0301|fyyfj96j0321|fyyfj96j0339|fyyfj96j0352|fyyfj96j0358|fyyfj96j0359|fyyfj96j0414|fyyfj96j0452|fyyfj96j0472|fyyfj96j0500|fyyfj96j0573|fyyfj97j0067|fyyfj97j0099|fyyfj97j0100|fyyfj97j0104|fyyfj97j0107|fyyfj97j0108|fyyfj97j0209|fyyfj97j0211|fyyfj97j0289|fyyfj97j0298|fyyfj97j0301|fyyfj97j0321|fyyfj97j0339|fyyfj97j0352|fyyfj97j0358|fyyfj97j0359|fyyfj97j0414|fyyfj97j0445|fyyfj97j0452|fyyfj97j0472|fyyfj97j0500|fyyfj97j0573|fyyfj98j0067|fyyfj98j0099|fyyfj98j0100|fyyfj98j0104|fyyfj98j0107|fyyfj98j0108|fyyfj98j0178|fyyfj98j0209|fyyfj98j0211|fyyfj98j0289|fyyfj98j0298|fyyfj98j0301|fyyfj98j0303|fyyfj98j0321|fyyfj98j0339|fyyfj98j0352|fyyfj98j0358|fyyfj98j0359|fyyfj98j0413|fyyfj98j0414|fyyfj98j0445|fyyfj98j0452|fyyfj98j0472|fyyfj98j0500|fyyfj98j0573|fyyfj99j0067|fyyfj99j0099|fyyfj99j0100|fyyfj99j0104|fyyfj99j0107|fyyfj99j0108|fyyfj99j0131|fyyfj99j0209|fyyfj99j0211|fyyfj99j0285|fyyfj99j0289|fyyfj99j0298|fyyfj99j0301|fyyfj99j0303|fyyfj99j0321|fyyfj99j0339|fyyfj99j0352|fyyfj99j0358|fyyfj99j0359|fyyfj99j0413|fyyfj99j0414|fyyfj99j0445|fyyfj99j0452|fyyfj99j0472|fyyfj99j0500|fyyfj99j0573|fyyfm01j0064|fyyfm01j0070|fyyfm01j0071|fyyfm01j0088|fyyfm01j0091|fyyfm01j0108|fyyfm01j0111|fyyfm01j0112|fyyfm01j0114|fyyfm01j0115|fyyfm01j0133|fyyfm01j0140|fyyfm01j0141|fyyfm01j0142|fyyfm01j0143|fyyfm01j0148|fyyfm01j0149|fyyfm01j0152|fyyfm01j0153|fyyfm01j0155|fyyfm01j0159|fyyfm01j0160|fyyfm01j0163|fyyfm01j0165|fyyfm01j0168|fyyfm01j0169|fyyfm01j0221|fyyfm01j0223|fyyfm01j0268|fyyfm01j0271|fyyfm01j0285|fyyfm01j0299|fyyfm01j0320|fyyfm01j0321|fyyfm01j0360|fyyfm01j0369|fyyfm01j0400|fyyfm01j0401|fyyfm01j0411|fyyfm01j0572|fyyfm01j0765|fyyfm02j0064|fyyfm02j0069|fyyfm02j0070|fyyfm02j0071|fyyfm02j0088|fyyfm02j0091|fyyfm02j0108|fyyfm02j0111|fyyfm02j0112|fyyfm02j0114|fyyfm02j0115|fyyfm02j0133|fyyfm02j0140|fyyfm02j0141|fyyfm02j0142|fyyfm02j0143|fyyfm02j0148|fyyfm02j0149|fyyfm02j0152|fyyfm02j0153|fyyfm02j0155|fyyfm02j0159|fyyfm02j0160|fyyfm02j0163|fyyfm02j0165|fyyfm02j0168|fyyfm02j0169|fyyfm02j0221|fyyfm02j0223|fyyfm02j0268|fyyfm02j0271|fyyfm02j0285|fyyfm02j0299|fyyfm02j0320|fyyfm02j0321|fyyfm02j0360|fyyfm02j0369|fyyfm02j0400|fyyfm02j0572|fyyfm02j0765|fyyfm03j0064|fyyfm03j0070|fyyfm03j0091|fyyfm03j0108|fyyfm03j0111|fyyfm03j0115|fyyfm03j0160|fyyfm03j0165|fyyfm03j0299|fyyfm03j0400|fyyfm03j0572|fyyfm04j0111|fyyfm51j0064|fyyfm51j0369|fyyfm52j0064|fyyfm52j0369|fyyfr88j0003|fyyfr89j0003|fyyff98j0071|fyyff98j0303|fyyff99j0029|fyyff99j0303|fyefj00j0112|fyefj00j0545|fyefj00j0546|fyefj00j0633|fyefj00j0634|fyefj00j0635|fyefj00j0636|fyefj00j0637|fyefj00j0649|fyefj00j0651|fyefj00j0652|fyefj00j0656|fyefj00j0657|fyefj00j0658|fyefj00j0659|fyefj00j0660|fyefj00j0685|fyefj00j0686|fyefj00j0688|fyefj00j0701|fyefj00j0702|fyefj00j0703|fyefj00j0715|fyefj00j0720|fyefj00j0721|fyefj00j0722|fyefj00j0724|fyefj00j0725|fyefj00j0726|fyefj00j0731|fyefj00j0751|fyefj00j0752|fyefj00j0756|fyefj00j0757|fyefj00j0758|fyefj00j0759|fyefj00j0761|fyefj00j0762|fyefj00j0763|fyefj00j0764|fyefj00j0768|fyefj00j0769|fyefj00j0785|fyefj00j0786|fyefj00j0789|fyefj00j0790|fyefj00j0793|fyefj00j0794|fyefj00j0803|fyefj00j0811|fyefj00j0821|fyefj00j0822|fyefj00j0823|fyefj00j0824|fyefj00j0825|fyefj00j0826|fyefj00j0827|fyefj00j0828|fyefj00j0829|fyefj00j0831|fyefj00j0832|fyefj00j0833|fyefj00j0838|fyefj00j0839|fyefj00j0840|fyefj00j0854|fyefj00j0855|fyefj00j0856|fyefj00j0859|fyefj00j0860|fyefj00j0861|fyefj00j0869|fyefj00j0870|fyefj00j0879|fyefj00j0887|fyefj00j0888|fyefj00j0889|fyefj00j0900|fyefj00j0901|fyefj00j0903|fyefj00j0904|fyefj00j0905|fyefj00j0959|fyefj00j0960|fyefj00j0961|fyefj00j1004|fyefj00j1005|fyefj00j1012|fyefj00j1013|fyefj00j1014|fyefj00j1015|fyefj00j1016|fyefj00j1017|fyefj00j1018|fyefj00j1019|fyefj00j1020|fyefj00j1021|fyefj00j1218|fyefj00j1219|fyefj00j1220|fyefj00j1221|fyefj00j1222|fyefj00j1811|fyefj00j1854|fyefj00j1855|fyefj00j1856|fyefj01j0707|fyefj02j0707|fyefj03j0707|fyefj66j0001|fyefj67j0001|fyefj68j0001|fyefj68j1064|fyefj69j0001|fyefj69j1064|fyefj70j0001|fyefj70j0859|fyefj70j1064|fyefj71j0001|fyefj71j1064|fyefj72j0001|fyefj72j1064|fyefj73j0001|fyefj73j1064|fyefj74j0001|fyefj74j1064|fyefj75j0001|fyefj75j1064|fyefj75j1092|fyefj76j0001|fyefj76j1064|fyefj76j1092|fyefj77j0001|fyefj77j1064|fyefj77j1092|fyefj78j0001|fyefj78j1064|fyefj78j1092|fyefj79j0001|fyefj79j1064|fyefj79j1092|fyefj80j0001|fyefj80j0859|fyefj80j1064|fyefj80j1077|fyefj80j1092|fyefj81j0001|fyefj81j1064|fyefj81j1077|fyefj81j1092|fyefj82j0001|fyefj82j1064|fyefj82j1092|fyefj83j0001|fyefj83j1064|fyefj83j1092|fyefj84j0001|fyefj84j1064|fyefj84j1092|fyefj85j0001|fyefj85j0356|fyefj85j1064|fyefj85j1092|fyefj86j0001|fyefj86j0356|fyefj86j1064|fyefj87j0001|fyefj87j0356|fyefj87j1064|fyefj88j0001|fyefj88j0356|fyefj88j1064|fyefj89j0001|fyefj89j0356|fyefj89j1064|fyefj89j1067|fyefj90j0001|fyefj90j0758|fyefj90j1021|fyefj90j1064|fyefj90j1067|fyefj91j0001|fyefj91j0758|fyefj91j0791|fyefj91j1021|fyefj91j1064|fyefj91j1067|fyefj91j1077|fyefj92j0001|fyefj92j0359|fyefj92j0678|fyefj92j0758|fyefj92j0791|fyefj92j0867|fyefj92j1021|fyefj92j1064|fyefj92j1077|fyefj93j0001|fyefj93j0359|fyefj93j0678|fyefj93j0758|fyefj93j0791|fyefj93j0867|fyefj93j1010|fyefj93j1021|fyefj93j1049|fyefj93j1064|fyefj93j1077|fyefj94j0001|fyefj94j0678|fyefj94j0758|fyefj94j0791|fyefj94j0867|fyefj94j1010|fyefj94j1021|fyefj94j1049|fyefj94j1064|fyefj94j1070|fyefj94j1077|fyefj94j1085|fyefj95j0001|fyefj95j0678|fyefj95j0758|fyefj95j0791|fyefj95j0867|fyefj95j0965|fyefj95j0966|fyefj95j1010|fyefj95j1011|fyefj95j1021|fyefj95j1055|fyefj95j1064|fyefj95j1069|fyefj95j1077|fyefj95j1085|fyefj95j1089|fyefj96j0001|fyefj96j0106|fyefj96j0671|fyefj96j0678|fyefj96j0758|fyefj96j0791|fyefj96j0814|fyefj96j0836|fyefj96j0867|fyefj96j0931|fyefj96j0965|fyefj96j0966|fyefj96j0976|fyefj96j1010|fyefj96j1021|fyefj96j1051|fyefj96j1055|fyefj96j1064|fyefj96j1068|fyefj96j1070|fyefj96j1077|fyefj96j1079|fyefj96j1081|fyefj96j1086|fyefj96j1088|fyefj96j1091|fyefj96j1093|fyefj96j1094|fyefj97j0001|fyefj97j0106|fyefj97j0584|fyefj97j0586|fyefj97j0671|fyefj97j0678|fyefj97j0758|fyefj97j0791|fyefj97j0814|fyefj97j0825|fyefj97j0836|fyefj97j0863|fyefj97j0865|fyefj97j0867|fyefj97j0914|fyefj97j0931|fyefj97j0952|fyefj97j0965|fyefj97j0966|fyefj97j0969|fyefj97j0971|fyefj97j0972|fyefj97j0976|fyefj97j0985|fyefj97j1010|fyefj97j1021|fyefj97j1051|fyefj97j1052|fyefj97j1055|fyefj97j1058|fyefj97j1059|fyefj97j1064|fyefj97j1068|fyefj97j1077|fyefj97j1079|fyefj97j1081|fyefj97j1086|fyefj97j1088|fyefj97j1095|fyefj98j0001|fyefj98j0243|fyefj98j0326|fyefj98j0329|fyefj98j0343|fyefj98j0344|fyefj98j0380|fyefj98j0472|fyefj98j0584|fyefj98j0586|fyefj98j0604|fyefj98j0671|fyefj98j0673|fyefj98j0676|fyefj98j0677|fyefj98j0678|fyefj98j0694|fyefj98j0758|fyefj98j0814|fyefj98j0825|fyefj98j0836|fyefj98j0863|fyefj98j0865|fyefj98j0867|fyefj98j0896|fyefj98j0898|fyefj98j0901|fyefj98j0906|fyefj98j0910|fyefj98j0913|fyefj98j0914|fyefj98j0922|fyefj98j0931|fyefj98j0934|fyefj98j0936|fyefj98j0951|fyefj98j0952|fyefj98j0963|fyefj98j0965|fyefj98j0966|fyefj98j0969|fyefj98j0971|fyefj98j0972|fyefj98j0974|fyefj98j0975|fyefj98j0976|fyefj98j0977|fyefj98j0978|fyefj98j0985|fyefj98j0992|fyefj98j1008|fyefj98j1009|fyefj98j1010|fyefj98j1011|fyefj98j1012|fyefj98j1019|fyefj98j1021|fyefj98j1028|fyefj98j1034|fyefj98j1039|fyefj98j1046|fyefj98j1047|fyefj98j1048|fyefj98j1054|fyefj98j1055|fyefj98j1064|fyefj98j1068|fyefj98j1077|fyefj98j1079|fyefj98j1080|fyefj98j1081|fyefj98j1082|fyefj98j1084|fyefj98j1087|fyefj98j1088|fyefj98j1090|fyefj99j0010|fyefj99j0188|fyefj99j0243|fyefj99j0268|fyefj99j0280|fyefj99j0301|fyefj99j0329|fyefj99j0343|fyefj99j0344|fyefj99j0380|fyefj99j0552|fyefj99j0573|fyefj99j0584|fyefj99j0586|fyefj99j0604|fyefj99j0671|fyefj99j0673|fyefj99j0676|fyefj99j0677|fyefj99j0678|fyefj99j0694|fyefj99j0722|fyefj99j0757|fyefj99j0758|fyefj99j0771|fyefj99j0772|fyefj99j0804|fyefj99j0806|fyefj99j0809|fyefj99j0814|fyefj99j0825|fyefj99j0836|fyefj99j0862|fyefj99j0863|fyefj99j0865|fyefj99j0866|fyefj99j0867|fyefj99j0875|fyefj99j0896|fyefj99j0898|fyefj99j0901|fyefj99j0906|fyefj99j0907|fyefj99j0908|fyefj99j0910|fyefj99j0912|fyefj99j0913|fyefj99j0914|fyefj99j0921|fyefj99j0922|fyefj99j0923|fyefj99j0931|fyefj99j0934|fyefj99j0936|fyefj99j0937|fyefj99j0949|fyefj99j0951|fyefj99j0952|fyefj99j0962|fyefj99j0963|fyefj99j0965|fyefj99j0966|fyefj99j0969|fyefj99j0971|fyefj99j0972|fyefj99j0974|fyefj99j0975|fyefj99j0976|fyefj99j0977|fyefj99j0978|fyefj99j0982|fyefj99j0985|fyefj99j0986|fyefj99j0988|fyefj99j0991|fyefj99j0992|fyefj99j0995|fyefj99j0997|fyefj99j0999|fyefj99j1003|fyefj99j1006|fyefj99j1008|fyefj99j1009|fyefj99j1010|fyefj99j1011|fyefj99j1016|fyefj99j1019|fyefj99j1020|fyefj99j1021|fyefj99j1024|fyefj99j1026|fyefj99j1028|fyefj99j1031|fyefj99j1033|fyefj99j1034|fyefj99j1036|fyefj99j1039|fyefj99j1042|fyefj99j1045|fyefj99j1046|fyefj99j1048|fyefj99j1053|fyefj99j1054|fyefj99j1055|fyefj99j1061|fyefj99j1062|fyefj99j1063|fyefj99j1064|fyefj99j1068|fyefj99j1072|fyefj99j1076|fyefj99j1077|fyefj99j1079|fyefj99j1080|fyefj99j1081|fyefj99j1083|fyefj99j1084|fyefj99j1087|fyefj99j1088|fyefm00j0113|fyefm01j0057|fyefm01j0088|fyefm01j0091|fyefm01j0101|fyefm01j0104|fyefm01j0107|fyefm01j0112|fyefm01j0379|fyefm02j0057|fyefm02j0101|fyefm02j0104|fyefm02j0107|fyefm02j0112|fyefm02j0379|fyefm98j0066|fyefm99j0066|fyefm99j0090|fyefm99j0093|fyefm99j0110|fyefm99j0165|fyefm99j0208|fyefm99j0209|fyefm99j0295|fyefm99j0401|fyefm99j0402|fyefm99j0907|fyefm99j1054|fyefn98j0015|fyefn98j0024|fyefn98j0030|fyefn99j0015|fyefn99j0024|fyefn99j0030|fyefr94j0559|fyefr95j0559|fyefr96j0559|fyefr97j0559|fyefr98j0559|fyefr99j0012|fyefr99j0559|fyefb01305|fyeff00j0170|fyeff00j0224|fyeff00j0227|fyeff00j0228|fyeff00j0229|fyeff00j0280|fyeff00j0281|fyeff00j0282|fyeff00j0283|fyeff00j0288|fyeff00j0289|fyeff00j0331|fyeff00j0332|fyeff00j0333|fyeff00j0334|fyeff00j0335|fyeff00j0336|fyeff00j0337|fyeff00j0338|fyeff00j0346|fyeff00j0347|fyeff00j0348|fyeff00j0349|fyeff00j0350|fyeff00j0351|fyeff00j0357|fyeff00j0358|fyeff00j0371|fyeff00j0372|fyeff00j0396|fyeff00j0397|fyeff00j0424|fyeff00j0425|fyeff01j0416|fyeff02j0416|fyeff78j0418|fyeff79j0418|fyeff79j1051|fyeff80j1051|fyeff81j1051|fyeff82j1051|fyeff83j1051|fyeff84j1051|fyeff85j1051|fyeff86j1051|fyeff87j1051|fyeff88j0422|fyeff89j0422|fyeff90j0422|fyeff90j0434|fyeff90j0440|fyeff91j0422|fyeff91j0434|fyeff91j0440|fyeff92j0440|fyeff93j0440|fyeff93j1045|fyeff93j1067|fyeff94j0392|fyeff94j0440|fyeff94j0443|fyeff94j1045|fyeff94j1067|fyeff95j0219|fyeff95j0392|fyeff95j0439|fyeff95j0440|fyeff95j0443|fyeff96j0053|fyeff96j0219|fyeff96j0392|fyeff96j0429|fyeff96j0434|fyeff96j0950|fyeff96j1019|fyeff96j1028|fyeff97j0053|fyeff97j0178|fyeff97j0191|fyeff97j0219|fyeff97j0221|fyeff97j0258|fyeff97j0324|fyeff97j0355|fyeff97j0370|fyeff97j0377|fyeff97j0392|fyeff97j0429|fyeff97j0434|fyeff97j0950|fyeff97j1019|fyeff98j0053|fyeff98j0065|fyeff98j0101|fyeff98j0144|fyeff98j0156|fyeff98j0178|fyeff98j0191|fyeff98j0193|fyeff98j0196|fyeff98j0197|fyeff98j0209|fyeff98j0210|fyeff98j0211|fyeff98j0214|fyeff98j0215|fyeff98j0218|fyeff98j0219|fyeff98j0221|fyeff98j0258|fyeff98j0260|fyeff98j0279|fyeff98j0284|fyeff98j0295|fyeff98j0296|fyeff98j0298|fyeff98j0324|fyeff98j0355|fyeff98j0370|fyeff98j0376|fyeff98j0379|fyeff98j0381|fyeff98j0392|fyeff98j0401|fyeff98j0404|fyeff98j0405|fyeff98j0407|fyeff98j0411|fyeff98j0418|fyeff98j0421|fyeff98j0423|fyeff98j0433|fyeff98j0436|fyeff98j0673|fyeff98j0896|fyeff98j0950|fyeff98j0985|fyeff98j1012|fyeff99j0053|fyeff99j0065|fyeff99j0152|fyeff99j0156|fyeff99j0159|fyeff99j0178|fyeff99j0191|fyeff99j0193|fyeff99j0196|fyeff99j0197|fyeff99j0209|fyeff99j0210|fyeff99j0211|fyeff99j0214|fyeff99j0215|fyeff99j0218|fyeff99j0219|fyeff99j0220|fyeff99j0221|fyeff99j0260|fyeff99j0279|fyeff99j0284|fyeff99j0291|fyeff99j0295|fyeff99j0296|fyeff99j0297|fyeff99j0298|fyeff99j0324|fyeff99j0339|fyeff99j0355|fyeff99j0370|fyeff99j0376|fyeff99j0379|fyeff99j0381|fyeff99j0392|fyeff99j0401|fyeff99j0404|fyeff99j0405|fyeff99j0407|fyeff99j0410|fyeff99j0411|fyeff99j0413|fyeff99j0414|fyeff99j0415|fyeff99j0418|fyeff99j0421|fyeff99j0423|fyeff99j0436|fyeff99j0673|fyeff99j0896|fyeff99j0950|fyeff99j0962|fyeff99j0985|fyeff99j1010|fyeff99j1012|fyeff99j1028|fyeff99j1090|fyeff99j1370|fayfm01j0148|fayfm01j0149|fayfm01j0155|fayfm02j0148|fayfm02j0149|fayfm02j0155|faefj00j0594|faefj00j0595|faefj00j0596|faefj00j0597|faefj01j0707|faefj02j0707|faefj03j0707|faefj90j1023|faefj91j1023|faefj92j1023|faefj94j1056|faefj95j1023|faefj95j1056|faefj96j1056|faefj98j1038|faefj99j1078|fdeff99j9001|fdeff99j9002|gyefj99j0005", + // A long case insensitive alternation. + "(?i:(zQPbMkNO|NNSPdvMi|iWuuSoAl|qbvKMimS|IecrXtPa|seTckYqt|NxnyHkgB|fIDlOgKb|UhlWIygH|OtNoJxHG|cUTkFVIV|mTgFIHjr|jQkoIDtE|PPMKxRXl|AwMfwVkQ|CQyMrTQJ|BzrqxVSi|nTpcWuhF|PertdywG|ZZDgCtXN|WWdDPyyE|uVtNQsKk|BdeCHvPZ|wshRnFlH|aOUIitIp|RxZeCdXT|CFZMslCj|AVBZRDxl|IzIGCnhw|ythYuWiz|oztXVXhl|VbLkwqQx|qvaUgyVC|VawUjPWC|ecloYJuj|boCLTdSU|uPrKeAZx|hrMWLWBq|JOnUNHRM|rYnujkPq|dDEdZhIj|DRrfvugG|yEGfDxVV|YMYdJWuP|PHUQZNWM|AmKNrLis|zTxndVfn|FPsHoJnc|EIulZTua|KlAPhdzg|ScHJJCLt|NtTfMzME|eMCwuFdo|SEpJVJbR|cdhXZeCx|sAVtBwRh|kVFEVcMI|jzJrxraA|tGLHTell|NNWoeSaw|DcOKSetX|UXZAJyka|THpMphDP|rizheevl|kDCBRidd|pCZZRqyu|pSygkitl|SwZGkAaW|wILOrfNX|QkwVOerj|kHOMxPDr|EwOVycJv|AJvtzQFS|yEOjKYYB|LizIINLL|JBRSsfcG|YPiUqqNl|IsdEbvee|MjEpGcBm|OxXZVgEQ|xClXGuxa|UzRCGFEb|buJbvfvA|IPZQxRet|oFYShsMc|oBHffuHO|bzzKrcBR|KAjzrGCl|IPUsAVls|OGMUMbIU|gyDccHuR|bjlalnDd|ZLWjeMna|fdsuIlxQ|dVXtiomV|XxedTjNg|XWMHlNoA|nnyqArQX|opfkWGhb|wYtnhdYb))", + // A long case insensitive alternation where each entry ends with ".*". + "(?i:(zQPbMkNO.*|NNSPdvMi.*|iWuuSoAl.*|qbvKMimS.*|IecrXtPa.*|seTckYqt.*|NxnyHkgB.*|fIDlOgKb.*|UhlWIygH.*|OtNoJxHG.*|cUTkFVIV.*|mTgFIHjr.*|jQkoIDtE.*|PPMKxRXl.*|AwMfwVkQ.*|CQyMrTQJ.*|BzrqxVSi.*|nTpcWuhF.*|PertdywG.*|ZZDgCtXN.*|WWdDPyyE.*|uVtNQsKk.*|BdeCHvPZ.*|wshRnFlH.*|aOUIitIp.*|RxZeCdXT.*|CFZMslCj.*|AVBZRDxl.*|IzIGCnhw.*|ythYuWiz.*|oztXVXhl.*|VbLkwqQx.*|qvaUgyVC.*|VawUjPWC.*|ecloYJuj.*|boCLTdSU.*|uPrKeAZx.*|hrMWLWBq.*|JOnUNHRM.*|rYnujkPq.*|dDEdZhIj.*|DRrfvugG.*|yEGfDxVV.*|YMYdJWuP.*|PHUQZNWM.*|AmKNrLis.*|zTxndVfn.*|FPsHoJnc.*|EIulZTua.*|KlAPhdzg.*|ScHJJCLt.*|NtTfMzME.*|eMCwuFdo.*|SEpJVJbR.*|cdhXZeCx.*|sAVtBwRh.*|kVFEVcMI.*|jzJrxraA.*|tGLHTell.*|NNWoeSaw.*|DcOKSetX.*|UXZAJyka.*|THpMphDP.*|rizheevl.*|kDCBRidd.*|pCZZRqyu.*|pSygkitl.*|SwZGkAaW.*|wILOrfNX.*|QkwVOerj.*|kHOMxPDr.*|EwOVycJv.*|AJvtzQFS.*|yEOjKYYB.*|LizIINLL.*|JBRSsfcG.*|YPiUqqNl.*|IsdEbvee.*|MjEpGcBm.*|OxXZVgEQ.*|xClXGuxa.*|UzRCGFEb.*|buJbvfvA.*|IPZQxRet.*|oFYShsMc.*|oBHffuHO.*|bzzKrcBR.*|KAjzrGCl.*|IPUsAVls.*|OGMUMbIU.*|gyDccHuR.*|bjlalnDd.*|ZLWjeMna.*|fdsuIlxQ.*|dVXtiomV.*|XxedTjNg.*|XWMHlNoA.*|nnyqArQX.*|opfkWGhb.*|wYtnhdYb.*))", + // A long case insensitive alternation where each entry starts with ".*". + "(?i:(.*zQPbMkNO|.*NNSPdvMi|.*iWuuSoAl|.*qbvKMimS|.*IecrXtPa|.*seTckYqt|.*NxnyHkgB|.*fIDlOgKb|.*UhlWIygH|.*OtNoJxHG|.*cUTkFVIV|.*mTgFIHjr|.*jQkoIDtE|.*PPMKxRXl|.*AwMfwVkQ|.*CQyMrTQJ|.*BzrqxVSi|.*nTpcWuhF|.*PertdywG|.*ZZDgCtXN|.*WWdDPyyE|.*uVtNQsKk|.*BdeCHvPZ|.*wshRnFlH|.*aOUIitIp|.*RxZeCdXT|.*CFZMslCj|.*AVBZRDxl|.*IzIGCnhw|.*ythYuWiz|.*oztXVXhl|.*VbLkwqQx|.*qvaUgyVC|.*VawUjPWC|.*ecloYJuj|.*boCLTdSU|.*uPrKeAZx|.*hrMWLWBq|.*JOnUNHRM|.*rYnujkPq|.*dDEdZhIj|.*DRrfvugG|.*yEGfDxVV|.*YMYdJWuP|.*PHUQZNWM|.*AmKNrLis|.*zTxndVfn|.*FPsHoJnc|.*EIulZTua|.*KlAPhdzg|.*ScHJJCLt|.*NtTfMzME|.*eMCwuFdo|.*SEpJVJbR|.*cdhXZeCx|.*sAVtBwRh|.*kVFEVcMI|.*jzJrxraA|.*tGLHTell|.*NNWoeSaw|.*DcOKSetX|.*UXZAJyka|.*THpMphDP|.*rizheevl|.*kDCBRidd|.*pCZZRqyu|.*pSygkitl|.*SwZGkAaW|.*wILOrfNX|.*QkwVOerj|.*kHOMxPDr|.*EwOVycJv|.*AJvtzQFS|.*yEOjKYYB|.*LizIINLL|.*JBRSsfcG|.*YPiUqqNl|.*IsdEbvee|.*MjEpGcBm|.*OxXZVgEQ|.*xClXGuxa|.*UzRCGFEb|.*buJbvfvA|.*IPZQxRet|.*oFYShsMc|.*oBHffuHO|.*bzzKrcBR|.*KAjzrGCl|.*IPUsAVls|.*OGMUMbIU|.*gyDccHuR|.*bjlalnDd|.*ZLWjeMna|.*fdsuIlxQ|.*dVXtiomV|.*XxedTjNg|.*XWMHlNoA|.*nnyqArQX|.*opfkWGhb|.*wYtnhdYb))", + // Quest ".?". + "fo.?", + "foo.?", + "f.?o", + ".*foo.?", + ".?foo.+", + "foo.?|bar", } + values = []string{ + "foo", " foo bar", "bar", "buzz\nbar", "bar foo", "bfoo", "\n", "\nfoo", "foo\n", "hello foo world", "hello foo\n world", "", + "FOO", "Foo", "OO", "Oo", "\nfoo\n", strings.Repeat("f", 20), "prometheus", "prometheus_api_v1", "prometheus_api_v1_foo", + "10.0.1.20", "10.0.2.10", "10.0.3.30", "10.0.4.40", + "foofoo0", "foofoo", - for _, c := range cases { - m, err := NewFastRegexMatcher(c.regex) - require.NoError(t, err) - require.Equal(t, c.expected, m.MatchString(c.value)) + // Values matching / not matching the test regexps on long alternations. + "zQPbMkNO", "zQPbMkNo", "jyyfj00j0061", "jyyfj00j006", "jyyfj00j00612", "NNSPdvMi", "NNSPdvMiXXX", "NNSPdvMixxx", "nnSPdvMi", "nnSPdvMiXXX", } +) + +func TestFastRegexMatcher_MatchString(t *testing.T) { + // Run the test both against a set of predefined values and a set of random ones. + testValues := append([]string{}, values...) + testValues = append(testValues, generateRandomValues()...) + + for _, r := range regexes { + r := r + for _, v := range testValues { + v := v + t.Run(readable(r)+` on "`+readable(v)+`"`, func(t *testing.T) { + t.Parallel() + m, err := NewFastRegexMatcher(r) + require.NoError(t, err) + re := regexp.MustCompile("^(?:" + r + ")$") + require.Equal(t, re.MatchString(v), m.MatchString(v)) + }) + } + } +} + +func readable(s string) string { + const maxReadableStringLen = 40 + if len(s) < maxReadableStringLen { + return s + } + return s[:maxReadableStringLen] + "..." } func TestOptimizeConcatRegex(t *testing.T) { @@ -85,6 +144,9 @@ func TestOptimizeConcatRegex(t *testing.T) { {regex: "(?i).*(?-i:abc)def", prefix: "", suffix: "", contains: "abc"}, {regex: ".*(?msU:abc).*", prefix: "", suffix: "", contains: "abc"}, {regex: "[aA]bc.*", prefix: "", suffix: "", contains: "bc"}, + {regex: "^5..$", prefix: "5", suffix: "", contains: ""}, + {regex: "^release.*", prefix: "release", suffix: "", contains: ""}, + {regex: "^env-[0-9]+laio[1]?[^0-9].*", prefix: "env-", suffix: "", contains: "laio"}, } for _, c := range cases { @@ -98,41 +160,906 @@ func TestOptimizeConcatRegex(t *testing.T) { } } -func BenchmarkFastRegexMatcher(b *testing.B) { - var ( - x = strings.Repeat("x", 50) - y = "foo" + x - z = x + "foo" - ) - regexes := []string{ - "foo", - "^foo", - "(foo|bar)", - "foo.*", - ".*foo", - "^.*foo$", - "^.+foo$", - ".*", - ".+", - "foo.+", - ".+foo", - ".*foo.*", - "(?i:foo)", - "(prometheus|api_prom)_api_v1_.+", - "((fo(bar))|.+foo)", - } - for _, r := range regexes { - r := r - b.Run(r, func(b *testing.B) { - m, err := NewFastRegexMatcher(r) - require.NoError(b, err) - b.ResetTimer() - for i := 0; i < b.N; i++ { - _ = m.MatchString(x) - _ = m.MatchString(y) - _ = m.MatchString(z) +// Refer to https://github.com/prometheus/prometheus/issues/2651. +func TestFindSetMatches(t *testing.T) { + for _, c := range []struct { + pattern string + expMatches []string + expCaseSensitive bool + }{ + // Single value, coming from a `bar=~"foo"` selector. + {"foo", []string{"foo"}, true}, + {"^foo", []string{"foo"}, true}, + {"^foo$", []string{"foo"}, true}, + // Simple sets alternates. + {"foo|bar|zz", []string{"foo", "bar", "zz"}, true}, + // Simple sets alternate and concat (bar|baz is parsed as "ba[rz]"). + {"foo|bar|baz", []string{"foo", "bar", "baz"}, true}, + // Simple sets alternate and concat and capture + {"foo|bar|baz|(zz)", []string{"foo", "bar", "baz", "zz"}, true}, + // Simple sets alternate and concat and alternates with empty matches + // parsed as b(ar|(?:)|uzz) where b(?:) means literal b. + {"bar|b|buzz", []string{"bar", "b", "buzz"}, true}, + // Skip nested capture groups. + {"^((bar|b|buzz))$", []string{"bar", "b", "buzz"}, true}, + // Skip outer anchors (it's enforced anyway at the root). + {"^(bar|b|buzz)$", []string{"bar", "b", "buzz"}, true}, + {"^(?:prod|production)$", []string{"prod", "production"}, true}, + // Do not optimize regexp with inner anchors. + {"(bar|b|b^uz$z)", nil, false}, + // Do not optimize regexp with empty string matcher. + {"^$|Running", nil, false}, + // Simple sets containing escaped characters. + {"fo\\.o|bar\\?|\\^baz", []string{"fo.o", "bar?", "^baz"}, true}, + // using charclass + {"[abc]d", []string{"ad", "bd", "cd"}, true}, + // high low charset different => A(B[CD]|EF)|BC[XY] + {"ABC|ABD|AEF|BCX|BCY", []string{"ABC", "ABD", "AEF", "BCX", "BCY"}, true}, + // triple concat + {"api_(v1|prom)_push", []string{"api_v1_push", "api_prom_push"}, true}, + // triple concat with multiple alternates + {"(api|rpc)_(v1|prom)_push", []string{"api_v1_push", "api_prom_push", "rpc_v1_push", "rpc_prom_push"}, true}, + {"(api|rpc)_(v1|prom)_(push|query)", []string{"api_v1_push", "api_v1_query", "api_prom_push", "api_prom_query", "rpc_v1_push", "rpc_v1_query", "rpc_prom_push", "rpc_prom_query"}, true}, + // class starting with "-" + {"[-1-2][a-c]", []string{"-a", "-b", "-c", "1a", "1b", "1c", "2a", "2b", "2c"}, true}, + {"[1^3]", []string{"1", "3", "^"}, true}, + // OpPlus with concat + {"(.+)/(foo|bar)", nil, false}, + // Simple sets containing special characters without escaping. + {"fo.o|bar?|^baz", nil, false}, + // case sensitive wrapper. + {"(?i)foo", []string{"FOO"}, false}, + // case sensitive wrapper on alternate. + {"(?i)foo|bar|baz", []string{"FOO", "BAR", "BAZ", "BAr", "BAz"}, false}, + // mixed case sensitivity. + {"(api|rpc)_(v1|prom)_((?i)push|query)", nil, false}, + // mixed case sensitivity concatenation only without capture group. + {"api_v1_(?i)push", nil, false}, + // mixed case sensitivity alternation only without capture group. + {"api|(?i)rpc", nil, false}, + // case sensitive after unsetting insensitivity. + {"rpc|(?i)(?-i)api", []string{"rpc", "api"}, true}, + // case sensitive after unsetting insensitivity in all alternation options. + {"(?i)((?-i)api|(?-i)rpc)", []string{"api", "rpc"}, true}, + // mixed case sensitivity after unsetting insensitivity. + {"(?i)rpc|(?-i)api", nil, false}, + // too high charset combination + {"(api|rpc)_[^0-9]", nil, false}, + // too many combinations + {"[a-z][a-z]", nil, false}, + } { + c := c + t.Run(c.pattern, func(t *testing.T) { + t.Parallel() + parsed, err := syntax.Parse(c.pattern, syntax.Perl) + require.NoError(t, err) + matches, actualCaseSensitive := findSetMatches(parsed) + require.Equal(t, c.expMatches, matches) + require.Equal(t, c.expCaseSensitive, actualCaseSensitive) + + if c.expCaseSensitive { + // When the regexp is case sensitive, we want to ensure that the + // set matches are maintained in the final matcher. + r, err := NewFastRegexMatcher(c.pattern) + require.NoError(t, err) + require.Equal(t, c.expMatches, r.SetMatches()) } }) - + } +} + +func TestFastRegexMatcher_SetMatches_ShouldReturnACopy(t *testing.T) { + m, err := NewFastRegexMatcher("a|b") + require.NoError(t, err) + require.Equal(t, []string{"a", "b"}, m.SetMatches()) + + // Manipulate the returned slice. + matches := m.SetMatches() + matches[0] = "xxx" + matches[1] = "yyy" + + // Ensure that if we call SetMatches() again we get the original one. + require.Equal(t, []string{"a", "b"}, m.SetMatches()) +} + +func BenchmarkFastRegexMatcher(b *testing.B) { + texts := generateRandomValues() + + for _, r := range regexes { + b.Run(getTestNameFromRegexp(r), func(b *testing.B) { + m, err := NewFastRegexMatcher(r) + require.NoError(b, err) + + b.ResetTimer() + for i := 0; i < b.N; i++ { + for _, text := range texts { + _ = m.MatchString(text) + } + } + }) + } +} + +func TestStringMatcherFromRegexp(t *testing.T) { + for _, c := range []struct { + pattern string + exp StringMatcher + }{ + {".*", anyStringWithoutNewlineMatcher{}}, + {".*?", anyStringWithoutNewlineMatcher{}}, + {"(?s:.*)", trueMatcher{}}, + {"(.*)", anyStringWithoutNewlineMatcher{}}, + {"^.*$", anyStringWithoutNewlineMatcher{}}, + {".+", &anyNonEmptyStringMatcher{matchNL: false}}, + {"(?s:.+)", &anyNonEmptyStringMatcher{matchNL: true}}, + {"^.+$", &anyNonEmptyStringMatcher{matchNL: false}}, + {"(.+)", &anyNonEmptyStringMatcher{matchNL: false}}, + {"", emptyStringMatcher{}}, + {"^$", emptyStringMatcher{}}, + {"^foo$", &equalStringMatcher{s: "foo", caseSensitive: true}}, + {"^(?i:foo)$", &equalStringMatcher{s: "FOO", caseSensitive: false}}, + {"^((?i:foo)|(bar))$", orStringMatcher([]StringMatcher{&equalStringMatcher{s: "FOO", caseSensitive: false}, &equalStringMatcher{s: "bar", caseSensitive: true}})}, + {`(?i:((foo|bar)))`, orStringMatcher([]StringMatcher{&equalStringMatcher{s: "FOO", caseSensitive: false}, &equalStringMatcher{s: "BAR", caseSensitive: false}})}, + {`(?i:((foo1|foo2|bar)))`, orStringMatcher([]StringMatcher{orStringMatcher([]StringMatcher{&equalStringMatcher{s: "FOO1", caseSensitive: false}, &equalStringMatcher{s: "FOO2", caseSensitive: false}}), &equalStringMatcher{s: "BAR", caseSensitive: false}})}, + {"^((?i:foo|oo)|(bar))$", orStringMatcher([]StringMatcher{&equalStringMatcher{s: "FOO", caseSensitive: false}, &equalStringMatcher{s: "OO", caseSensitive: false}, &equalStringMatcher{s: "bar", caseSensitive: true}})}, + {"(?i:(foo1|foo2|bar))", orStringMatcher([]StringMatcher{orStringMatcher([]StringMatcher{&equalStringMatcher{s: "FOO1", caseSensitive: false}, &equalStringMatcher{s: "FOO2", caseSensitive: false}}), &equalStringMatcher{s: "BAR", caseSensitive: false}})}, + {".*foo.*", &containsStringMatcher{substrings: []string{"foo"}, left: anyStringWithoutNewlineMatcher{}, right: anyStringWithoutNewlineMatcher{}}}, + {"(.*)foo.*", &containsStringMatcher{substrings: []string{"foo"}, left: anyStringWithoutNewlineMatcher{}, right: anyStringWithoutNewlineMatcher{}}}, + {"(.*)foo(.*)", &containsStringMatcher{substrings: []string{"foo"}, left: anyStringWithoutNewlineMatcher{}, right: anyStringWithoutNewlineMatcher{}}}, + {"(.+)foo(.*)", &containsStringMatcher{substrings: []string{"foo"}, left: &anyNonEmptyStringMatcher{matchNL: false}, right: anyStringWithoutNewlineMatcher{}}}, + {"^.+foo.+", &containsStringMatcher{substrings: []string{"foo"}, left: &anyNonEmptyStringMatcher{matchNL: false}, right: &anyNonEmptyStringMatcher{matchNL: false}}}, + {"^(.*)(foo)(.*)$", &containsStringMatcher{substrings: []string{"foo"}, left: anyStringWithoutNewlineMatcher{}, right: anyStringWithoutNewlineMatcher{}}}, + {"^(.*)(foo|foobar)(.*)$", &containsStringMatcher{substrings: []string{"foo", "foobar"}, left: anyStringWithoutNewlineMatcher{}, right: anyStringWithoutNewlineMatcher{}}}, + {"^(.*)(foo|foobar)(.+)$", &containsStringMatcher{substrings: []string{"foo", "foobar"}, left: anyStringWithoutNewlineMatcher{}, right: &anyNonEmptyStringMatcher{matchNL: false}}}, + {"^(.*)(bar|b|buzz)(.+)$", &containsStringMatcher{substrings: []string{"bar", "b", "buzz"}, left: anyStringWithoutNewlineMatcher{}, right: &anyNonEmptyStringMatcher{matchNL: false}}}, + {"10\\.0\\.(1|2)\\.+", nil}, + {"10\\.0\\.(1|2).+", &containsStringMatcher{substrings: []string{"10.0.1", "10.0.2"}, left: nil, right: &anyNonEmptyStringMatcher{matchNL: false}}}, + {"^.+foo", &literalSuffixStringMatcher{left: &anyNonEmptyStringMatcher{}, suffix: "foo", suffixCaseSensitive: true}}, + {"foo-.*$", &literalPrefixStringMatcher{prefix: "foo-", prefixCaseSensitive: true, right: anyStringWithoutNewlineMatcher{}}}, + {"(prometheus|api_prom)_api_v1_.+", &containsStringMatcher{substrings: []string{"prometheus_api_v1_", "api_prom_api_v1_"}, left: nil, right: &anyNonEmptyStringMatcher{matchNL: false}}}, + {"^((.*)(bar|b|buzz)(.+)|foo)$", orStringMatcher([]StringMatcher{&containsStringMatcher{substrings: []string{"bar", "b", "buzz"}, left: anyStringWithoutNewlineMatcher{}, right: &anyNonEmptyStringMatcher{matchNL: false}}, &equalStringMatcher{s: "foo", caseSensitive: true}})}, + {"((fo(bar))|.+foo)", orStringMatcher([]StringMatcher{orStringMatcher([]StringMatcher{&equalStringMatcher{s: "fobar", caseSensitive: true}}), &literalSuffixStringMatcher{suffix: "foo", suffixCaseSensitive: true, left: &anyNonEmptyStringMatcher{matchNL: false}}})}, + {"(.+)/(gateway|cortex-gw|cortex-gw-internal)", &containsStringMatcher{substrings: []string{"/gateway", "/cortex-gw", "/cortex-gw-internal"}, left: &anyNonEmptyStringMatcher{matchNL: false}, right: nil}}, + // we don't support case insensitive matching for contains. + // This is because there's no strings.IndexOfFold function. + // We can revisit later if this is really popular by using strings.ToUpper. + {"^(.*)((?i)foo|foobar)(.*)$", nil}, + {"(api|rpc)_(v1|prom)_((?i)push|query)", nil}, + {"[a-z][a-z]", nil}, + {"[1^3]", nil}, + {".*foo.*bar.*", nil}, + {`\d*`, nil}, + {".", nil}, + {"/|/bar.*", &literalPrefixStringMatcher{prefix: "/", prefixCaseSensitive: true, right: orStringMatcher{emptyStringMatcher{}, &literalPrefixStringMatcher{prefix: "bar", prefixCaseSensitive: true, right: anyStringWithoutNewlineMatcher{}}}}}, + // This one is not supported because `stringMatcherFromRegexp` is not reentrant for syntax.OpConcat. + // It would make the code too complex to handle it. + {"(.+)/(foo.*|bar$)", nil}, + // Case sensitive alternate with same literal prefix and .* suffix. + {"(xyz-016a-ixb-dp.*|xyz-016a-ixb-op.*)", &literalPrefixStringMatcher{prefix: "xyz-016a-ixb-", prefixCaseSensitive: true, right: orStringMatcher{&literalPrefixStringMatcher{prefix: "dp", prefixCaseSensitive: true, right: anyStringWithoutNewlineMatcher{}}, &literalPrefixStringMatcher{prefix: "op", prefixCaseSensitive: true, right: anyStringWithoutNewlineMatcher{}}}}}, + // Case insensitive alternate with same literal prefix and .* suffix. + {"(?i:(xyz-016a-ixb-dp.*|xyz-016a-ixb-op.*))", &literalPrefixStringMatcher{prefix: "XYZ-016A-IXB-", prefixCaseSensitive: false, right: orStringMatcher{&literalPrefixStringMatcher{prefix: "DP", prefixCaseSensitive: false, right: anyStringWithoutNewlineMatcher{}}, &literalPrefixStringMatcher{prefix: "OP", prefixCaseSensitive: false, right: anyStringWithoutNewlineMatcher{}}}}}, + {"(?i)(xyz-016a-ixb-dp.*|xyz-016a-ixb-op.*)", &literalPrefixStringMatcher{prefix: "XYZ-016A-IXB-", prefixCaseSensitive: false, right: orStringMatcher{&literalPrefixStringMatcher{prefix: "DP", prefixCaseSensitive: false, right: anyStringWithoutNewlineMatcher{}}, &literalPrefixStringMatcher{prefix: "OP", prefixCaseSensitive: false, right: anyStringWithoutNewlineMatcher{}}}}}, + // Concatenated variable length selectors are not supported. + {"foo.*.*", nil}, + {"foo.+.+", nil}, + {".*.*foo", nil}, + {".+.+foo", nil}, + {"aaa.?.?", nil}, + {"aaa.?.*", nil}, + // Regexps with ".?". + {"ext.?|xfs", orStringMatcher{&literalPrefixStringMatcher{prefix: "ext", prefixCaseSensitive: true, right: &zeroOrOneCharacterStringMatcher{matchNL: false}}, &equalStringMatcher{s: "xfs", caseSensitive: true}}}, + {"(?s)(ext.?|xfs)", orStringMatcher{&literalPrefixStringMatcher{prefix: "ext", prefixCaseSensitive: true, right: &zeroOrOneCharacterStringMatcher{matchNL: true}}, &equalStringMatcher{s: "xfs", caseSensitive: true}}}, + {"foo.?", &literalPrefixStringMatcher{prefix: "foo", prefixCaseSensitive: true, right: &zeroOrOneCharacterStringMatcher{matchNL: false}}}, + {"f.?o", nil}, + } { + c := c + t.Run(c.pattern, func(t *testing.T) { + t.Parallel() + parsed, err := syntax.Parse(c.pattern, syntax.Perl) + require.NoError(t, err) + matches := stringMatcherFromRegexp(parsed) + require.Equal(t, c.exp, matches) + }) + } +} + +func TestStringMatcherFromRegexp_LiteralPrefix(t *testing.T) { + for _, c := range []struct { + pattern string + expectedLiteralPrefixMatchers int + expectedMatches []string + expectedNotMatches []string + }{ + // Case sensitive. + { + pattern: "(xyz-016a-ixb-dp.*|xyz-016a-ixb-op.*)", + expectedLiteralPrefixMatchers: 3, + expectedMatches: []string{"xyz-016a-ixb-dp", "xyz-016a-ixb-dpXXX", "xyz-016a-ixb-op", "xyz-016a-ixb-opXXX"}, + expectedNotMatches: []string{"XYZ-016a-ixb-dp", "xyz-016a-ixb-d", "XYZ-016a-ixb-op", "xyz-016a-ixb-o", "xyz", "dp", "xyz-016a-ixb-dp\n"}, + }, + + // Case insensitive. + { + pattern: "(?i)(xyz-016a-ixb-dp.*|xyz-016a-ixb-op.*)", + expectedLiteralPrefixMatchers: 3, + expectedMatches: []string{"xyz-016a-ixb-dp", "XYZ-016a-ixb-dpXXX", "xyz-016a-ixb-op", "XYZ-016a-ixb-opXXX"}, + expectedNotMatches: []string{"xyz-016a-ixb-d", "xyz", "dp", "xyz-016a-ixb-dp\n"}, + }, + + // Nested literal prefixes, case sensitive. + { + pattern: "(xyz-(aaa-(111.*)|bbb-(222.*)))|(xyz-(aaa-(333.*)|bbb-(444.*)))", + expectedLiteralPrefixMatchers: 10, + expectedMatches: []string{"xyz-aaa-111", "xyz-aaa-111XXX", "xyz-aaa-333", "xyz-aaa-333XXX", "xyz-bbb-222", "xyz-bbb-222XXX", "xyz-bbb-444", "xyz-bbb-444XXX"}, + expectedNotMatches: []string{"XYZ-aaa-111", "xyz-aaa-11", "xyz-aaa-222", "xyz-bbb-111"}, + }, + + // Nested literal prefixes, case insensitive. + { + pattern: "(?i)(xyz-(aaa-(111.*)|bbb-(222.*)))|(xyz-(aaa-(333.*)|bbb-(444.*)))", + expectedLiteralPrefixMatchers: 10, + expectedMatches: []string{"xyz-aaa-111", "XYZ-aaa-111XXX", "xyz-aaa-333", "xyz-AAA-333XXX", "xyz-bbb-222", "xyz-BBB-222XXX", "XYZ-bbb-444", "xyz-bbb-444XXX"}, + expectedNotMatches: []string{"xyz-aaa-11", "xyz-aaa-222", "xyz-bbb-111"}, + }, + + // Mixed case sensitivity. + { + pattern: "(xyz-((?i)(aaa.*|bbb.*)))", + expectedLiteralPrefixMatchers: 3, + expectedMatches: []string{"xyz-aaa", "xyz-AAA", "xyz-aaaXXX", "xyz-AAAXXX", "xyz-bbb", "xyz-BBBXXX"}, + expectedNotMatches: []string{"XYZ-aaa", "xyz-aa", "yz-aaa", "aaa"}, + }, + } { + t.Run(c.pattern, func(t *testing.T) { + parsed, err := syntax.Parse(c.pattern, syntax.Perl) + require.NoError(t, err) + + matcher := stringMatcherFromRegexp(parsed) + require.NotNil(t, matcher) + + re := regexp.MustCompile("^" + c.pattern + "$") + + // Pre-condition check: ensure it contains literalPrefixStringMatcher. + numPrefixMatchers := 0 + visitStringMatcher(matcher, func(matcher StringMatcher) { + if _, ok := matcher.(*literalPrefixStringMatcher); ok { + numPrefixMatchers++ + } + }) + + require.Equal(t, c.expectedLiteralPrefixMatchers, numPrefixMatchers) + + for _, value := range c.expectedMatches { + require.Truef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Truef(t, re.MatchString(value), "Value: %s", value) + } + + for _, value := range c.expectedNotMatches { + require.Falsef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Falsef(t, re.MatchString(value), "Value: %s", value) + } + }) + } +} + +func TestStringMatcherFromRegexp_LiteralSuffix(t *testing.T) { + for _, c := range []struct { + pattern string + expectedLiteralSuffixMatchers int + expectedMatches []string + expectedNotMatches []string + }{ + // Case sensitive. + { + pattern: "(.*xyz-016a-ixb-dp|.*xyz-016a-ixb-op)", + expectedLiteralSuffixMatchers: 2, + expectedMatches: []string{"xyz-016a-ixb-dp", "XXXxyz-016a-ixb-dp", "xyz-016a-ixb-op", "XXXxyz-016a-ixb-op"}, + expectedNotMatches: []string{"XYZ-016a-ixb-dp", "yz-016a-ixb-dp", "XYZ-016a-ixb-op", "xyz-016a-ixb-o", "xyz", "dp", "\nxyz-016a-ixb-dp"}, + }, + + // Case insensitive. + { + pattern: "(?i)(.*xyz-016a-ixb-dp|.*xyz-016a-ixb-op)", + expectedLiteralSuffixMatchers: 2, + expectedMatches: []string{"xyz-016a-ixb-dp", "XYZ-016a-ixb-dp", "XXXxyz-016a-ixb-dp", "XyZ-016a-ixb-op", "XXXxyz-016a-ixb-op"}, + expectedNotMatches: []string{"yz-016a-ixb-dp", "xyz-016a-ixb-o", "xyz", "dp", "\nxyz-016a-ixb-dp"}, + }, + + // Nested literal suffixes, case sensitive. + { + pattern: "(.*aaa|.*bbb(.*ccc|.*ddd))", + expectedLiteralSuffixMatchers: 3, + expectedMatches: []string{"aaa", "XXXaaa", "bbbccc", "XXXbbbccc", "XXXbbbXXXccc", "bbbddd", "bbbddd", "XXXbbbddd", "XXXbbbXXXddd", "bbbXXXccc", "aaabbbccc", "aaabbbddd"}, + expectedNotMatches: []string{"AAA", "aa", "Xaa", "BBBCCC", "bb", "Xbb", "bbccc", "bbbcc", "bbbdd"}, + }, + + // Mixed case sensitivity. + { + pattern: "(.*aaa|.*bbb((?i)(.*ccc|.*ddd)))", + expectedLiteralSuffixMatchers: 3, + expectedMatches: []string{"aaa", "XXXaaa", "bbbccc", "bbbCCC", "bbbXXXCCC", "bbbddd", "bbbDDD", "bbbXXXddd", "bbbXXXDDD"}, + expectedNotMatches: []string{"AAA", "XXXAAA", "BBBccc", "BBBCCC", "aaaBBB"}, + }, + } { + t.Run(c.pattern, func(t *testing.T) { + parsed, err := syntax.Parse(c.pattern, syntax.Perl) + require.NoError(t, err) + + matcher := stringMatcherFromRegexp(parsed) + require.NotNil(t, matcher) + + re := regexp.MustCompile("^" + c.pattern + "$") + + // Pre-condition check: ensure it contains literalSuffixStringMatcher. + numSuffixMatchers := 0 + visitStringMatcher(matcher, func(matcher StringMatcher) { + if _, ok := matcher.(*literalSuffixStringMatcher); ok { + numSuffixMatchers++ + } + }) + + require.Equal(t, c.expectedLiteralSuffixMatchers, numSuffixMatchers) + + for _, value := range c.expectedMatches { + require.Truef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Truef(t, re.MatchString(value), "Value: %s", value) + } + + for _, value := range c.expectedNotMatches { + require.Falsef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Falsef(t, re.MatchString(value), "Value: %s", value) + } + }) + } +} + +func TestStringMatcherFromRegexp_Quest(t *testing.T) { + for _, c := range []struct { + pattern string + expectedZeroOrOneMatchers int + expectedMatches []string + expectedNotMatches []string + }{ + // Not match newline. + { + pattern: "test.?", + expectedZeroOrOneMatchers: 1, + expectedMatches: []string{"test", "test!"}, + expectedNotMatches: []string{"test\n", "tes", "test!!"}, + }, + { + pattern: ".?test", + expectedZeroOrOneMatchers: 1, + expectedMatches: []string{"test", "!test"}, + expectedNotMatches: []string{"\ntest", "tes", "test!"}, + }, + { + pattern: "(aaa.?|bbb.?)", + expectedZeroOrOneMatchers: 2, + expectedMatches: []string{"aaa", "aaaX", "bbb", "bbbX"}, + expectedNotMatches: []string{"aa", "aaaXX", "aaa\n", "bb", "bbbXX", "bbb\n"}, + }, + { + pattern: ".*aaa.?", + expectedZeroOrOneMatchers: 1, + expectedMatches: []string{"aaa", "Xaaa", "aaaX", "XXXaaa", "XXXaaaX"}, + expectedNotMatches: []string{"aa", "aaaXX", "XXXaaaXXX", "XXXaaa\n"}, + }, + + // Match newline. + { + pattern: "(?s)test.?", + expectedZeroOrOneMatchers: 1, + expectedMatches: []string{"test", "test!", "test\n"}, + expectedNotMatches: []string{"tes", "test!!", "test\n\n"}, + }, + + // Mixed flags (a part matches newline another doesn't). + { + pattern: "(aaa.?|((?s).?bbb.+))", + expectedZeroOrOneMatchers: 2, + expectedMatches: []string{"aaa", "aaaX", "bbbX", "XbbbX", "bbbXXX", "\nbbbX"}, + expectedNotMatches: []string{"aa", "aaa\n", "Xbbb", "\nbbb"}, + }, + } { + t.Run(c.pattern, func(t *testing.T) { + parsed, err := syntax.Parse(c.pattern, syntax.Perl) + require.NoError(t, err) + + matcher := stringMatcherFromRegexp(parsed) + require.NotNil(t, matcher) + + re := regexp.MustCompile("^" + c.pattern + "$") + + // Pre-condition check: ensure it contains zeroOrOneCharacterStringMatcher. + numZeroOrOneMatchers := 0 + visitStringMatcher(matcher, func(matcher StringMatcher) { + if _, ok := matcher.(*zeroOrOneCharacterStringMatcher); ok { + numZeroOrOneMatchers++ + } + }) + + require.Equal(t, c.expectedZeroOrOneMatchers, numZeroOrOneMatchers) + + for _, value := range c.expectedMatches { + require.Truef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Truef(t, re.MatchString(value), "Value: %s", value) + } + + for _, value := range c.expectedNotMatches { + require.Falsef(t, matcher.Matches(value), "Value: %s", value) + + // Ensure the golang regexp engine would return the same. + require.Falsef(t, re.MatchString(value), "Value: %s", value) + } + }) + } +} + +func randString(randGenerator *rand.Rand, length int) string { + b := make([]rune, length) + for i := range b { + b[i] = asciiRunes[randGenerator.Intn(len(asciiRunes))] + } + return string(b) +} + +func randStrings(randGenerator *rand.Rand, many, length int) []string { + out := make([]string, 0, many) + for i := 0; i < many; i++ { + out = append(out, randString(randGenerator, length)) + } + return out +} + +func TestOptimizeEqualStringMatchers(t *testing.T) { + tests := map[string]struct { + input StringMatcher + expectedValues []string + expectedCaseSensitive bool + }{ + "should skip optimization on orStringMatcher with containsStringMatcher": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + &containsStringMatcher{substrings: []string{"a", "b", "c"}}, + }, + expectedValues: nil, + }, + "should run optimization on orStringMatcher with equalStringMatcher and same case sensitivity": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + &equalStringMatcher{s: "bar", caseSensitive: true}, + &equalStringMatcher{s: "baz", caseSensitive: true}, + }, + expectedValues: []string{"FOO", "bar", "baz"}, + expectedCaseSensitive: true, + }, + "should skip optimization on orStringMatcher with equalStringMatcher but different case sensitivity": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + &equalStringMatcher{s: "bar", caseSensitive: false}, + &equalStringMatcher{s: "baz", caseSensitive: true}, + }, + expectedValues: nil, + }, + "should run optimization on orStringMatcher with nested orStringMatcher and equalStringMatcher, and same case sensitivity": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + orStringMatcher{ + &equalStringMatcher{s: "bar", caseSensitive: true}, + &equalStringMatcher{s: "xxx", caseSensitive: true}, + }, + &equalStringMatcher{s: "baz", caseSensitive: true}, + }, + expectedValues: []string{"FOO", "bar", "xxx", "baz"}, + expectedCaseSensitive: true, + }, + "should skip optimization on orStringMatcher with nested orStringMatcher and equalStringMatcher, but different case sensitivity": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + orStringMatcher{ + // Case sensitivity is different within items at the same level. + &equalStringMatcher{s: "bar", caseSensitive: true}, + &equalStringMatcher{s: "xxx", caseSensitive: false}, + }, + &equalStringMatcher{s: "baz", caseSensitive: true}, + }, + expectedValues: nil, + }, + "should skip optimization on orStringMatcher with nested orStringMatcher and equalStringMatcher, but different case sensitivity in the nested one": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: true}, + // Case sensitivity is different between the parent and child. + orStringMatcher{ + &equalStringMatcher{s: "bar", caseSensitive: false}, + &equalStringMatcher{s: "xxx", caseSensitive: false}, + }, + &equalStringMatcher{s: "baz", caseSensitive: true}, + }, + expectedValues: nil, + }, + "should return unchanged values on few case insensitive matchers": { + input: orStringMatcher{ + &equalStringMatcher{s: "FOO", caseSensitive: false}, + orStringMatcher{ + &equalStringMatcher{s: "bAr", caseSensitive: false}, + }, + &equalStringMatcher{s: "baZ", caseSensitive: false}, + }, + expectedValues: []string{"FOO", "bAr", "baZ"}, + expectedCaseSensitive: false, + }, + } + + for testName, testData := range tests { + t.Run(testName, func(t *testing.T) { + actualMatcher := optimizeEqualStringMatchers(testData.input, 0) + + if testData.expectedValues == nil { + require.IsType(t, testData.input, actualMatcher) + } else { + require.IsType(t, &equalMultiStringSliceMatcher{}, actualMatcher) + require.Equal(t, testData.expectedValues, actualMatcher.(*equalMultiStringSliceMatcher).values) + require.Equal(t, testData.expectedCaseSensitive, actualMatcher.(*equalMultiStringSliceMatcher).caseSensitive) + } + }) + } +} + +func TestNewEqualMultiStringMatcher(t *testing.T) { + tests := map[string]struct { + values []string + caseSensitive bool + expectedValuesMap map[string]struct{} + expectedValuesList []string + }{ + "few case sensitive values": { + values: []string{"a", "B"}, + caseSensitive: true, + expectedValuesList: []string{"a", "B"}, + }, + "few case insensitive values": { + values: []string{"a", "B"}, + caseSensitive: false, + expectedValuesList: []string{"a", "B"}, + }, + "many case sensitive values": { + values: []string{"a", "B", "c", "D", "e", "F", "g", "H", "i", "L", "m", "N", "o", "P", "q", "r"}, + caseSensitive: true, + expectedValuesMap: map[string]struct{}{"a": {}, "B": {}, "c": {}, "D": {}, "e": {}, "F": {}, "g": {}, "H": {}, "i": {}, "L": {}, "m": {}, "N": {}, "o": {}, "P": {}, "q": {}, "r": {}}, + }, + "many case insensitive values": { + values: []string{"a", "B", "c", "D", "e", "F", "g", "H", "i", "L", "m", "N", "o", "P", "q", "r"}, + caseSensitive: false, + expectedValuesMap: map[string]struct{}{"a": {}, "b": {}, "c": {}, "d": {}, "e": {}, "f": {}, "g": {}, "h": {}, "i": {}, "l": {}, "m": {}, "n": {}, "o": {}, "p": {}, "q": {}, "r": {}}, + }, + } + + for testName, testData := range tests { + t.Run(testName, func(t *testing.T) { + matcher := newEqualMultiStringMatcher(testData.caseSensitive, len(testData.values)) + for _, v := range testData.values { + matcher.add(v) + } + if testData.expectedValuesMap != nil { + require.IsType(t, &equalMultiStringMapMatcher{}, matcher) + require.Equal(t, testData.expectedValuesMap, matcher.(*equalMultiStringMapMatcher).values) + require.Equal(t, testData.caseSensitive, matcher.(*equalMultiStringMapMatcher).caseSensitive) + } + if testData.expectedValuesList != nil { + require.IsType(t, &equalMultiStringSliceMatcher{}, matcher) + require.Equal(t, testData.expectedValuesList, matcher.(*equalMultiStringSliceMatcher).values) + require.Equal(t, testData.caseSensitive, matcher.(*equalMultiStringSliceMatcher).caseSensitive) + } + }) + } +} + +func TestEqualMultiStringMatcher_Matches(t *testing.T) { + tests := map[string]struct { + values []string + caseSensitive bool + expectedMatches []string + expectedNotMatches []string + }{ + "few case sensitive values": { + values: []string{"a", "B"}, + caseSensitive: true, + expectedMatches: []string{"a", "B"}, + expectedNotMatches: []string{"A", "b"}, + }, + "few case insensitive values": { + values: []string{"a", "B"}, + caseSensitive: false, + expectedMatches: []string{"a", "A", "b", "B"}, + expectedNotMatches: []string{"c", "C"}, + }, + "many case sensitive values": { + values: []string{"a", "B", "c", "D", "e", "F", "g", "H", "i", "L", "m", "N", "o", "P", "q", "r"}, + caseSensitive: true, + expectedMatches: []string{"a", "B"}, + expectedNotMatches: []string{"A", "b"}, + }, + "many case insensitive values": { + values: []string{"a", "B", "c", "D", "e", "F", "g", "H", "i", "L", "m", "N", "o", "P", "q", "r"}, + caseSensitive: false, + expectedMatches: []string{"a", "A", "b", "B"}, + expectedNotMatches: []string{"x", "X"}, + }, + } + + for testName, testData := range tests { + t.Run(testName, func(t *testing.T) { + matcher := newEqualMultiStringMatcher(testData.caseSensitive, len(testData.values)) + for _, v := range testData.values { + matcher.add(v) + } + + for _, v := range testData.expectedMatches { + require.True(t, matcher.Matches(v), "value: %s", v) + } + for _, v := range testData.expectedNotMatches { + require.False(t, matcher.Matches(v), "value: %s", v) + } + }) + } +} + +func TestFindEqualStringMatchers(t *testing.T) { + type match struct { + s string + caseSensitive bool + } + + // Utility to call findEqualStringMatchers() and collect all callback invocations. + findEqualStringMatchersAndCollectMatches := func(input StringMatcher) (matches []match, ok bool) { + ok = findEqualStringMatchers(input, func(matcher *equalStringMatcher) bool { + matches = append(matches, match{matcher.s, matcher.caseSensitive}) + return true + }) + return + } + + t.Run("empty matcher", func(t *testing.T) { + actualMatches, actualOk := findEqualStringMatchersAndCollectMatches(emptyStringMatcher{}) + require.False(t, actualOk) + require.Empty(t, actualMatches) + }) + + t.Run("concat of literal matchers (case sensitive)", func(t *testing.T) { + actualMatches, actualOk := findEqualStringMatchersAndCollectMatches( + orStringMatcher{ + &equalStringMatcher{s: "test-1", caseSensitive: true}, + &equalStringMatcher{s: "test-2", caseSensitive: true}, + }, + ) + + require.True(t, actualOk) + require.Equal(t, []match{{"test-1", true}, {"test-2", true}}, actualMatches) + }) + + t.Run("concat of literal matchers (case insensitive)", func(t *testing.T) { + actualMatches, actualOk := findEqualStringMatchersAndCollectMatches( + orStringMatcher{ + &equalStringMatcher{s: "test-1", caseSensitive: false}, + &equalStringMatcher{s: "test-2", caseSensitive: false}, + }, + ) + + require.True(t, actualOk) + require.Equal(t, []match{{"test-1", false}, {"test-2", false}}, actualMatches) + }) + + t.Run("concat of literal matchers (mixed case)", func(t *testing.T) { + actualMatches, actualOk := findEqualStringMatchersAndCollectMatches( + orStringMatcher{ + &equalStringMatcher{s: "test-1", caseSensitive: false}, + &equalStringMatcher{s: "test-2", caseSensitive: true}, + }, + ) + + require.True(t, actualOk) + require.Equal(t, []match{{"test-1", false}, {"test-2", true}}, actualMatches) + }) +} + +// This benchmark is used to find a good threshold to use to apply the optimization +// done by optimizeEqualStringMatchers(). +func BenchmarkOptimizeEqualStringMatchers(b *testing.B) { + randGenerator := rand.New(rand.NewSource(time.Now().UnixNano())) + + // Generate variable lengths random texts to match against. + texts := append([]string{}, randStrings(randGenerator, 10, 10)...) + texts = append(texts, randStrings(randGenerator, 5, 30)...) + texts = append(texts, randStrings(randGenerator, 1, 100)...) + + for numAlternations := 2; numAlternations <= 256; numAlternations *= 2 { + for _, caseSensitive := range []bool{true, false} { + b.Run(fmt.Sprintf("alternations: %d case sensitive: %t", numAlternations, caseSensitive), func(b *testing.B) { + // Generate a regex with the expected number of alternations. + re := strings.Join(randStrings(randGenerator, numAlternations, 10), "|") + if !caseSensitive { + re = "(?i:(" + re + "))" + } + + parsed, err := syntax.Parse(re, syntax.Perl) + require.NoError(b, err) + + unoptimized := stringMatcherFromRegexpInternal(parsed) + require.IsType(b, orStringMatcher{}, unoptimized) + + optimized := optimizeEqualStringMatchers(unoptimized, 0) + if numAlternations < minEqualMultiStringMatcherMapThreshold { + require.IsType(b, &equalMultiStringSliceMatcher{}, optimized) + } else { + require.IsType(b, &equalMultiStringMapMatcher{}, optimized) + } + + b.Run("without optimizeEqualStringMatchers()", func(b *testing.B) { + for n := 0; n < b.N; n++ { + for _, t := range texts { + unoptimized.Matches(t) + } + } + }) + + b.Run("with optimizeEqualStringMatchers()", func(b *testing.B) { + for n := 0; n < b.N; n++ { + for _, t := range texts { + optimized.Matches(t) + } + } + }) + }) + } + } +} + +func TestZeroOrOneCharacterStringMatcher(t *testing.T) { + matcher := &zeroOrOneCharacterStringMatcher{matchNL: true} + require.True(t, matcher.Matches("")) + require.True(t, matcher.Matches("x")) + require.True(t, matcher.Matches("\n")) + require.False(t, matcher.Matches("xx")) + require.False(t, matcher.Matches("\n\n")) + + matcher = &zeroOrOneCharacterStringMatcher{matchNL: false} + require.True(t, matcher.Matches("")) + require.True(t, matcher.Matches("x")) + require.False(t, matcher.Matches("\n")) + require.False(t, matcher.Matches("xx")) + require.False(t, matcher.Matches("\n\n")) +} + +func TestLiteralPrefixStringMatcher(t *testing.T) { + m := &literalPrefixStringMatcher{prefix: "mar", prefixCaseSensitive: true, right: &emptyStringMatcher{}} + require.True(t, m.Matches("mar")) + require.False(t, m.Matches("marco")) + require.False(t, m.Matches("ma")) + require.False(t, m.Matches("mAr")) + + m = &literalPrefixStringMatcher{prefix: "mar", prefixCaseSensitive: false, right: &emptyStringMatcher{}} + require.True(t, m.Matches("mar")) + require.False(t, m.Matches("marco")) + require.False(t, m.Matches("ma")) + require.True(t, m.Matches("mAr")) + + m = &literalPrefixStringMatcher{prefix: "mar", prefixCaseSensitive: true, right: &equalStringMatcher{s: "co", caseSensitive: false}} + require.True(t, m.Matches("marco")) + require.True(t, m.Matches("marCO")) + require.False(t, m.Matches("MARco")) + require.False(t, m.Matches("mar")) + require.False(t, m.Matches("marcopracucci")) +} + +func TestLiteralSuffixStringMatcher(t *testing.T) { + m := &literalSuffixStringMatcher{left: &emptyStringMatcher{}, suffix: "co", suffixCaseSensitive: true} + require.True(t, m.Matches("co")) + require.False(t, m.Matches("marco")) + require.False(t, m.Matches("coo")) + require.False(t, m.Matches("Co")) + + m = &literalSuffixStringMatcher{left: &emptyStringMatcher{}, suffix: "co", suffixCaseSensitive: false} + require.True(t, m.Matches("co")) + require.False(t, m.Matches("marco")) + require.False(t, m.Matches("coo")) + require.True(t, m.Matches("Co")) + + m = &literalSuffixStringMatcher{left: &equalStringMatcher{s: "mar", caseSensitive: false}, suffix: "co", suffixCaseSensitive: true} + require.True(t, m.Matches("marco")) + require.True(t, m.Matches("MARco")) + require.False(t, m.Matches("marCO")) + require.False(t, m.Matches("mar")) + require.False(t, m.Matches("marcopracucci")) + + m = &literalSuffixStringMatcher{left: &equalStringMatcher{s: "mar", caseSensitive: false}, suffix: "co", suffixCaseSensitive: false} + require.True(t, m.Matches("marco")) + require.True(t, m.Matches("MARco")) + require.True(t, m.Matches("marCO")) + require.False(t, m.Matches("mar")) + require.False(t, m.Matches("marcopracucci")) +} + +func TestHasPrefixCaseInsensitive(t *testing.T) { + require.True(t, hasPrefixCaseInsensitive("marco", "mar")) + require.True(t, hasPrefixCaseInsensitive("mArco", "mar")) + require.True(t, hasPrefixCaseInsensitive("marco", "MaR")) + require.True(t, hasPrefixCaseInsensitive("marco", "marco")) + require.True(t, hasPrefixCaseInsensitive("mArco", "marco")) + + require.False(t, hasPrefixCaseInsensitive("marco", "a")) + require.False(t, hasPrefixCaseInsensitive("marco", "abcdefghi")) +} + +func TestHasSuffixCaseInsensitive(t *testing.T) { + require.True(t, hasSuffixCaseInsensitive("marco", "rco")) + require.True(t, hasSuffixCaseInsensitive("marco", "RcO")) + require.True(t, hasSuffixCaseInsensitive("marco", "marco")) + require.False(t, hasSuffixCaseInsensitive("marco", "a")) + require.False(t, hasSuffixCaseInsensitive("marco", "abcdefghi")) +} + +func getTestNameFromRegexp(re string) string { + if len(re) > 32 { + return re[:32] + } + return re +} + +func generateRandomValues() []string { + // Init the random seed with a constant, so that it doesn't change between runs. + randGenerator := rand.New(rand.NewSource(1)) + + // Generate variable lengths random texts to match against. + texts := append([]string{}, randStrings(randGenerator, 10, 10)...) + texts = append(texts, randStrings(randGenerator, 5, 30)...) + texts = append(texts, randStrings(randGenerator, 1, 100)...) + texts = append(texts, "foo"+randString(randGenerator, 50)) + texts = append(texts, randString(randGenerator, 50)+"foo") + + return texts +} + +func visitStringMatcher(matcher StringMatcher, callback func(matcher StringMatcher)) { + callback(matcher) + + switch casted := matcher.(type) { + case *containsStringMatcher: + if casted.left != nil { + visitStringMatcher(casted.left, callback) + } + if casted.right != nil { + visitStringMatcher(casted.right, callback) + } + + case *literalPrefixStringMatcher: + visitStringMatcher(casted.right, callback) + + case *literalSuffixStringMatcher: + visitStringMatcher(casted.left, callback) + + case orStringMatcher: + for _, entry := range casted { + visitStringMatcher(entry, callback) + } + + // No nested matchers for the following ones. + case emptyStringMatcher: + case *equalStringMatcher: + case *equalMultiStringSliceMatcher: + case *equalMultiStringMapMatcher: + case anyStringWithoutNewlineMatcher: + case *anyNonEmptyStringMatcher: + case trueMatcher: } } diff --git a/model/rulefmt/testdata/bad_field.bad.yaml b/model/rulefmt/testdata/bad_field.bad.yaml index d85eab1e5f..729bbadfbc 100644 --- a/model/rulefmt/testdata/bad_field.bad.yaml +++ b/model/rulefmt/testdata/bad_field.bad.yaml @@ -6,4 +6,4 @@ groups: labels: instance: localhost annotation: - summary: annonations is written without s above + summary: annotations is written without s above diff --git a/notifier/notifier.go b/notifier/notifier.go index ff8b05aa1e..4cf376aa05 100644 --- a/notifier/notifier.go +++ b/notifier/notifier.go @@ -595,7 +595,7 @@ func labelsToOpenAPILabelSet(modelLabelSet labels.Labels) models.LabelSet { } func (n *Manager) sendOne(ctx context.Context, c *http.Client, url string, b []byte) error { - req, err := http.NewRequest("POST", url, bytes.NewReader(b)) + req, err := http.NewRequest(http.MethodPost, url, bytes.NewReader(b)) if err != nil { return err } diff --git a/promql/engine.go b/promql/engine.go index 68e0502907..b8a8ea0959 100644 --- a/promql/engine.go +++ b/promql/engine.go @@ -115,6 +115,12 @@ func (e ErrStorage) Error() string { return e.Err.Error() } +// QueryEngine defines the interface for the *promql.Engine, so it can be replaced, wrapped or mocked. +type QueryEngine interface { + NewInstantQuery(ctx context.Context, q storage.Queryable, opts QueryOpts, qs string, ts time.Time) (Query, error) + NewRangeQuery(ctx context.Context, q storage.Queryable, opts QueryOpts, qs string, start, end time.Time, interval time.Duration) (Query, error) +} + // QueryLogger is an interface that can be used to log all the queries logged // by the engine. type QueryLogger interface { @@ -1061,8 +1067,6 @@ func (ev *evaluator) Eval(expr parser.Expr) (v parser.Value, ws annotations.Anno // EvalSeriesHelper stores extra information about a series. type EvalSeriesHelper struct { - // The grouping key used by aggregation. - groupingKey uint64 // Used to map left-hand to right-hand in binary operations. signature string } @@ -1075,8 +1079,6 @@ type EvalNodeHelper struct { Out Vector // Caches. - // label_*. - Dmn map[uint64]labels.Labels // funcHistogramQuantile for classic histograms. signatureToMetricWithBuckets map[string]*metricWithBuckets @@ -1196,6 +1198,9 @@ func (ev *evaluator) rangeEval(prepSeries func(labels.Labels, *EvalSeriesHelper) if prepSeries != nil { bufHelpers[i] = append(bufHelpers[i], seriesHelpers[i][si]) } + // Don't add histogram size here because we only + // copy the pointer above, not the whole + // histogram. ev.currentSamples++ if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) @@ -1221,7 +1226,6 @@ func (ev *evaluator) rangeEval(prepSeries func(labels.Labels, *EvalSeriesHelper) if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } - ev.samplesStats.UpdatePeak(ev.currentSamples) // If this could be an instant query, shortcut so as not to change sort order. if ev.endTimestamp == ev.startTimestamp { @@ -1253,17 +1257,7 @@ func (ev *evaluator) rangeEval(prepSeries func(labels.Labels, *EvalSeriesHelper) } else { ss = seriesAndTimestamp{Series{Metric: sample.Metric}, ts} } - if sample.H == nil { - if ss.Floats == nil { - ss.Floats = getFPointSlice(numSteps) - } - ss.Floats = append(ss.Floats, FPoint{T: ts, F: sample.F}) - } else { - if ss.Histograms == nil { - ss.Histograms = getHPointSlice(numSteps) - } - ss.Histograms = append(ss.Histograms, HPoint{T: ts, H: sample.H}) - } + addToSeries(&ss.Series, enh.Ts, sample.F, sample.H, numSteps) seriess[h] = ss } } @@ -1285,6 +1279,116 @@ func (ev *evaluator) rangeEval(prepSeries func(labels.Labels, *EvalSeriesHelper) return mat, warnings } +func (ev *evaluator) rangeEvalAgg(aggExpr *parser.AggregateExpr, sortedGrouping []string, inputMatrix Matrix, param float64) (Matrix, annotations.Annotations) { + // Keep a copy of the original point slice so that it can be returned to the pool. + origMatrix := slices.Clone(inputMatrix) + defer func() { + for _, s := range origMatrix { + putFPointSlice(s.Floats) + putHPointSlice(s.Histograms) + } + }() + + var warnings annotations.Annotations + + enh := &EvalNodeHelper{} + tempNumSamples := ev.currentSamples + + // Create a mapping from input series to output groups. + buf := make([]byte, 0, 1024) + groupToResultIndex := make(map[uint64]int) + seriesToResult := make([]int, len(inputMatrix)) + var result Matrix + + groupCount := 0 + for si, series := range inputMatrix { + var groupingKey uint64 + groupingKey, buf = generateGroupingKey(series.Metric, sortedGrouping, aggExpr.Without, buf) + index, ok := groupToResultIndex[groupingKey] + // Add a new group if it doesn't exist. + if !ok { + if aggExpr.Op != parser.TOPK && aggExpr.Op != parser.BOTTOMK { + m := generateGroupingLabels(enh, series.Metric, aggExpr.Without, sortedGrouping) + result = append(result, Series{Metric: m}) + } + index = groupCount + groupToResultIndex[groupingKey] = index + groupCount++ + } + seriesToResult[si] = index + } + groups := make([]groupedAggregation, groupCount) + + var k int + var seriess map[uint64]Series + switch aggExpr.Op { + case parser.TOPK, parser.BOTTOMK: + if !convertibleToInt64(param) { + ev.errorf("Scalar value %v overflows int64", param) + } + k = int(param) + if k > len(inputMatrix) { + k = len(inputMatrix) + } + if k < 1 { + return nil, warnings + } + seriess = make(map[uint64]Series, len(inputMatrix)) // Output series by series hash. + case parser.QUANTILE: + if math.IsNaN(param) || param < 0 || param > 1 { + warnings.Add(annotations.NewInvalidQuantileWarning(param, aggExpr.Param.PositionRange())) + } + } + + for ts := ev.startTimestamp; ts <= ev.endTimestamp; ts += ev.interval { + if err := contextDone(ev.ctx, "expression evaluation"); err != nil { + ev.error(err) + } + // Reset number of samples in memory after each timestamp. + ev.currentSamples = tempNumSamples + + // Make the function call. + enh.Ts = ts + var ws annotations.Annotations + switch aggExpr.Op { + case parser.TOPK, parser.BOTTOMK: + result, ws = ev.aggregationK(aggExpr, k, inputMatrix, seriesToResult, groups, enh, seriess) + // If this could be an instant query, shortcut so as not to change sort order. + if ev.endTimestamp == ev.startTimestamp { + return result, ws + } + default: + ws = ev.aggregation(aggExpr, param, inputMatrix, result, seriesToResult, groups, enh) + } + + warnings.Merge(ws) + + if ev.currentSamples > ev.maxSamples { + ev.error(ErrTooManySamples(env)) + } + } + + // Assemble the output matrix. By the time we get here we know we don't have too many samples. + switch aggExpr.Op { + case parser.TOPK, parser.BOTTOMK: + result = make(Matrix, 0, len(seriess)) + for _, ss := range seriess { + result = append(result, ss) + } + default: + // Remove empty result rows. + dst := 0 + for _, series := range result { + if len(series.Floats) > 0 || len(series.Histograms) > 0 { + result[dst] = series + dst++ + } + } + result = result[:dst] + } + return result, warnings +} + // evalSubquery evaluates given SubqueryExpr and returns an equivalent // evaluated MatrixSelector in its place. Note that the Name and LabelMatchers are not set. func (ev *evaluator) evalSubquery(subq *parser.SubqueryExpr) (*parser.MatrixSelector, int, annotations.Annotations) { @@ -1337,28 +1441,44 @@ func (ev *evaluator) eval(expr parser.Expr) (parser.Value, annotations.Annotatio sortedGrouping := e.Grouping slices.Sort(sortedGrouping) - // Prepare a function to initialise series helpers with the grouping key. - buf := make([]byte, 0, 1024) - initSeries := func(series labels.Labels, h *EvalSeriesHelper) { - h.groupingKey, buf = generateGroupingKey(series, sortedGrouping, e.Without, buf) - } - unwrapParenExpr(&e.Param) param := unwrapStepInvariantExpr(e.Param) unwrapParenExpr(¶m) - if s, ok := param.(*parser.StringLiteral); ok { - return ev.rangeEval(initSeries, func(v []parser.Value, sh [][]EvalSeriesHelper, enh *EvalNodeHelper) (Vector, annotations.Annotations) { - return ev.aggregation(e, sortedGrouping, s.Val, v[0].(Vector), sh[0], enh) + + if e.Op == parser.COUNT_VALUES { + valueLabel := param.(*parser.StringLiteral) + if !model.LabelName(valueLabel.Val).IsValid() { + ev.errorf("invalid label name %q", valueLabel) + } + if !e.Without { + sortedGrouping = append(sortedGrouping, valueLabel.Val) + slices.Sort(sortedGrouping) + } + return ev.rangeEval(nil, func(v []parser.Value, _ [][]EvalSeriesHelper, enh *EvalNodeHelper) (Vector, annotations.Annotations) { + return ev.aggregationCountValues(e, sortedGrouping, valueLabel.Val, v[0].(Vector), enh) }, e.Expr) } - return ev.rangeEval(initSeries, func(v []parser.Value, sh [][]EvalSeriesHelper, enh *EvalNodeHelper) (Vector, annotations.Annotations) { - var param float64 - if e.Param != nil { - param = v[0].(Vector)[0].F - } - return ev.aggregation(e, sortedGrouping, param, v[1].(Vector), sh[1], enh) - }, e.Param, e.Expr) + var warnings annotations.Annotations + originalNumSamples := ev.currentSamples + // param is the number k for topk/bottomk, or q for quantile. + var fParam float64 + if param != nil { + val, ws := ev.eval(param) + warnings.Merge(ws) + fParam = val.(Matrix)[0].Floats[0].F + } + // Now fetch the data to be aggregated. + val, ws := ev.eval(e.Expr) + warnings.Merge(ws) + inputMatrix := val.(Matrix) + + result, ws := ev.rangeEvalAgg(e, sortedGrouping, inputMatrix, fParam) + warnings.Merge(ws) + ev.currentSamples = originalNumSamples + result.TotalSamples() + ev.samplesStats.UpdatePeak(ev.currentSamples) + + return result, warnings case *parser.Call: call := FunctionCalls[e.Func.Name] @@ -1540,13 +1660,12 @@ func (ev *evaluator) eval(expr parser.Expr) (parser.Value, annotations.Annotatio histSamples := totalHPointSize(ss.Histograms) if len(ss.Floats)+histSamples > 0 { - if ev.currentSamples+len(ss.Floats)+histSamples <= ev.maxSamples { - mat = append(mat, ss) - prevSS = &mat[len(mat)-1] - ev.currentSamples += len(ss.Floats) + histSamples - } else { + if ev.currentSamples+len(ss.Floats)+histSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } + mat = append(mat, ss) + prevSS = &mat[len(mat)-1] + ev.currentSamples += len(ss.Floats) + histSamples } ev.samplesStats.UpdatePeak(ev.currentSamples) @@ -1709,26 +1828,28 @@ func (ev *evaluator) eval(expr parser.Expr) (parser.Value, annotations.Annotatio step++ _, f, h, ok := ev.vectorSelectorSingle(it, e, ts) if ok { - if ev.currentSamples < ev.maxSamples { - if h == nil { - if ss.Floats == nil { - ss.Floats = reuseOrGetFPointSlices(prevSS, numSteps) - } - ss.Floats = append(ss.Floats, FPoint{F: f, T: ts}) - ev.currentSamples++ - ev.samplesStats.IncrementSamplesAtStep(step, 1) - } else { - if ss.Histograms == nil { - ss.Histograms = reuseOrGetHPointSlices(prevSS, numSteps) - } - point := HPoint{H: h, T: ts} - ss.Histograms = append(ss.Histograms, point) - histSize := point.size() - ev.currentSamples += histSize - ev.samplesStats.IncrementSamplesAtStep(step, int64(histSize)) + if h == nil { + ev.currentSamples++ + ev.samplesStats.IncrementSamplesAtStep(step, 1) + if ev.currentSamples > ev.maxSamples { + ev.error(ErrTooManySamples(env)) } + if ss.Floats == nil { + ss.Floats = reuseOrGetFPointSlices(prevSS, numSteps) + } + ss.Floats = append(ss.Floats, FPoint{F: f, T: ts}) } else { - ev.error(ErrTooManySamples(env)) + point := HPoint{H: h, T: ts} + histSize := point.size() + ev.currentSamples += histSize + ev.samplesStats.IncrementSamplesAtStep(step, int64(histSize)) + if ev.currentSamples > ev.maxSamples { + ev.error(ErrTooManySamples(env)) + } + if ss.Histograms == nil { + ss.Histograms = reuseOrGetHPointSlices(prevSS, numSteps) + } + ss.Histograms = append(ss.Histograms, point) } } } @@ -1856,7 +1977,7 @@ func (ev *evaluator) eval(expr parser.Expr) (parser.Value, annotations.Annotatio panic(fmt.Errorf("unhandled expression of type: %T", expr)) } -// reuseOrGetFPointSlices reuses the space from previous slice to create new slice if the former has lots of room. +// reuseOrGetHPointSlices reuses the space from previous slice to create new slice if the former has lots of room. // The previous slices capacity is adjusted so when it is re-used from the pool it doesn't overflow into the new one. func reuseOrGetHPointSlices(prevSS *Series, numSteps int) (r []HPoint) { if prevSS != nil && cap(prevSS.Histograms)-2*len(prevSS.Histograms) > 0 { @@ -2168,10 +2289,10 @@ loop: histograms = histograms[:n] continue loop } - if ev.currentSamples >= ev.maxSamples { + ev.currentSamples += histograms[n].size() + if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } - ev.currentSamples += histograms[n].size() } case chunkenc.ValFloat: t, f := buf.At() @@ -2180,10 +2301,10 @@ loop: } // Values in the buffer are guaranteed to be smaller than maxt. if t >= mintFloats { - if ev.currentSamples >= ev.maxSamples { + ev.currentSamples++ + if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } - ev.currentSamples++ if floats == nil { floats = getFPointSlice(16) } @@ -2211,22 +2332,22 @@ loop: histograms = histograms[:n] break } - if ev.currentSamples >= ev.maxSamples { + ev.currentSamples += histograms[n].size() + if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } - ev.currentSamples += histograms[n].size() case chunkenc.ValFloat: t, f := it.At() if t == maxt && !value.IsStaleNaN(f) { - if ev.currentSamples >= ev.maxSamples { + ev.currentSamples++ + if ev.currentSamples > ev.maxSamples { ev.error(ErrTooManySamples(env)) } if floats == nil { floats = getFPointSlice(16) } floats = append(floats, FPoint{T: t, F: f}) - ev.currentSamples++ } } ev.samplesStats.UpdatePeak(ev.currentSamples) @@ -2607,171 +2728,85 @@ func vectorElemBinop(op parser.ItemType, lhs, rhs float64, hlhs, hrhs *histogram } type groupedAggregation struct { + seen bool // Was this output groups seen in the input at this timestamp. hasFloat bool // Has at least 1 float64 sample aggregated. hasHistogram bool // Has at least 1 histogram sample aggregated. - labels labels.Labels floatValue float64 histogramValue *histogram.FloatHistogram floatMean float64 - histogramMean *histogram.FloatHistogram groupCount int heap vectorByValueHeap - reverseHeap vectorByReverseValueHeap } -// aggregation evaluates an aggregation operation on a Vector. The provided grouping labels -// must be sorted. -func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, param interface{}, vec Vector, seriesHelper []EvalSeriesHelper, enh *EvalNodeHelper) (Vector, annotations.Annotations) { +// aggregation evaluates sum, avg, count, stdvar, stddev or quantile at one timestep on inputMatrix. +// These functions produce one output series for each group specified in the expression, with just the labels from `by(...)`. +// outputMatrix should be already populated with grouping labels; groups is one-to-one with outputMatrix. +// seriesToResult maps inputMatrix indexes to outputMatrix indexes. +func (ev *evaluator) aggregation(e *parser.AggregateExpr, q float64, inputMatrix, outputMatrix Matrix, seriesToResult []int, groups []groupedAggregation, enh *EvalNodeHelper) annotations.Annotations { op := e.Op - without := e.Without var annos annotations.Annotations - result := map[uint64]*groupedAggregation{} - orderedResult := []*groupedAggregation{} - var k int64 - if op == parser.TOPK || op == parser.BOTTOMK { - f := param.(float64) - if !convertibleToInt64(f) { - ev.errorf("Scalar value %v overflows int64", f) - } - k = int64(f) - if k < 1 { - return Vector{}, annos - } - } - var q float64 - if op == parser.QUANTILE { - q = param.(float64) - } - var valueLabel string - var recomputeGroupingKey bool - if op == parser.COUNT_VALUES { - valueLabel = param.(string) - if !model.LabelName(valueLabel).IsValid() { - ev.errorf("invalid label name %q", valueLabel) - } - if !without { - // We're changing the grouping labels so we have to ensure they're still sorted - // and we have to flag to recompute the grouping key. Considering the count_values() - // operator is less frequently used than other aggregations, we're fine having to - // re-compute the grouping key on each step for this case. - grouping = append(grouping, valueLabel) - slices.Sort(grouping) - recomputeGroupingKey = true - } + for i := range groups { + groups[i].seen = false } - var buf []byte - for si, s := range vec { - metric := s.Metric - - if op == parser.COUNT_VALUES { - enh.resetBuilder(metric) - enh.lb.Set(valueLabel, strconv.FormatFloat(s.F, 'f', -1, 64)) - metric = enh.lb.Labels() - - // We've changed the metric so we have to recompute the grouping key. - recomputeGroupingKey = true - } - - // We can use the pre-computed grouping key unless grouping labels have changed. - var groupingKey uint64 - if !recomputeGroupingKey { - groupingKey = seriesHelper[si].groupingKey - } else { - groupingKey, buf = generateGroupingKey(metric, grouping, without, buf) - } - - group, ok := result[groupingKey] - // Add a new group if it doesn't exist. + for si := range inputMatrix { + f, h, ok := ev.nextValues(enh.Ts, &inputMatrix[si]) if !ok { - var m labels.Labels - enh.resetBuilder(metric) - switch { - case without: - enh.lb.Del(grouping...) - enh.lb.Del(labels.MetricName) - m = enh.lb.Labels() - case len(grouping) > 0: - enh.lb.Keep(grouping...) - m = enh.lb.Labels() - default: - m = labels.EmptyLabels() - } - newAgg := &groupedAggregation{ - labels: m, - floatValue: s.F, - floatMean: s.F, + continue + } + + group := &groups[seriesToResult[si]] + // Initialize this group if it's the first time we've seen it. + if !group.seen { + *group = groupedAggregation{ + seen: true, + floatValue: f, + floatMean: f, groupCount: 1, } - switch { - case s.H == nil: - newAgg.hasFloat = true - case op == parser.SUM: - newAgg.histogramValue = s.H.Copy() - newAgg.hasHistogram = true - case op == parser.AVG: - newAgg.histogramMean = s.H.Copy() - newAgg.hasHistogram = true - case op == parser.STDVAR || op == parser.STDDEV: - newAgg.groupCount = 0 - } - - result[groupingKey] = newAgg - orderedResult = append(orderedResult, newAgg) - - inputVecLen := int64(len(vec)) - resultSize := k - switch { - case k > inputVecLen: - resultSize = inputVecLen - case k == 0: - resultSize = 1 - } switch op { + case parser.SUM, parser.AVG: + if h == nil { + group.hasFloat = true + } else { + group.histogramValue = h.Copy() + group.hasHistogram = true + } case parser.STDVAR, parser.STDDEV: - result[groupingKey].floatValue = 0 - case parser.TOPK, parser.QUANTILE: - result[groupingKey].heap = make(vectorByValueHeap, 1, resultSize) - result[groupingKey].heap[0] = Sample{ - F: s.F, - Metric: s.Metric, - } - case parser.BOTTOMK: - result[groupingKey].reverseHeap = make(vectorByReverseValueHeap, 1, resultSize) - result[groupingKey].reverseHeap[0] = Sample{ - F: s.F, - Metric: s.Metric, - } + group.floatValue = 0 + case parser.QUANTILE: + group.heap = make(vectorByValueHeap, 1) + group.heap[0] = Sample{F: f} case parser.GROUP: - result[groupingKey].floatValue = 1 + group.floatValue = 1 } continue } switch op { case parser.SUM: - if s.H != nil { + if h != nil { group.hasHistogram = true if group.histogramValue != nil { - group.histogramValue.Add(s.H) + group.histogramValue.Add(h) } // Otherwise the aggregation contained floats // previously and will be invalid anyway. No // point in copying the histogram in that case. } else { group.hasFloat = true - group.floatValue += s.F + group.floatValue += f } case parser.AVG: group.groupCount++ - if s.H != nil { + if h != nil { group.hasHistogram = true - if group.histogramMean != nil { - left := s.H.Copy().Div(float64(group.groupCount)) - right := group.histogramMean.Copy().Div(float64(group.groupCount)) + if group.histogramValue != nil { + left := h.Copy().Div(float64(group.groupCount)) + right := group.histogramValue.Copy().Div(float64(group.groupCount)) toAdd := left.Sub(right) - group.histogramMean.Add(toAdd) + group.histogramValue.Add(toAdd) } // Otherwise the aggregation contained floats // previously and will be invalid anyway. No @@ -2779,13 +2814,13 @@ func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, par } else { group.hasFloat = true if math.IsInf(group.floatMean, 0) { - if math.IsInf(s.F, 0) && (group.floatMean > 0) == (s.F > 0) { + if math.IsInf(f, 0) && (group.floatMean > 0) == (f > 0) { // The `floatMean` and `s.F` values are `Inf` of the same sign. They // can't be subtracted, but the value of `floatMean` is correct // already. break } - if !math.IsInf(s.F, 0) && !math.IsNaN(s.F) { + if !math.IsInf(f, 0) && !math.IsNaN(f) { // At this stage, the mean is an infinite. If the added // value is neither an Inf or a Nan, we can keep that mean // value. @@ -2796,81 +2831,48 @@ func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, par } } // Divide each side of the `-` by `group.groupCount` to avoid float64 overflows. - group.floatMean += s.F/float64(group.groupCount) - group.floatMean/float64(group.groupCount) + group.floatMean += f/float64(group.groupCount) - group.floatMean/float64(group.groupCount) } case parser.GROUP: // Do nothing. Required to avoid the panic in `default:` below. case parser.MAX: - if group.floatValue < s.F || math.IsNaN(group.floatValue) { - group.floatValue = s.F + if group.floatValue < f || math.IsNaN(group.floatValue) { + group.floatValue = f } case parser.MIN: - if group.floatValue > s.F || math.IsNaN(group.floatValue) { - group.floatValue = s.F + if group.floatValue > f || math.IsNaN(group.floatValue) { + group.floatValue = f } - case parser.COUNT, parser.COUNT_VALUES: + case parser.COUNT: group.groupCount++ case parser.STDVAR, parser.STDDEV: - if s.H == nil { // Ignore native histograms. + if h == nil { // Ignore native histograms. group.groupCount++ - delta := s.F - group.floatMean + delta := f - group.floatMean group.floatMean += delta / float64(group.groupCount) - group.floatValue += delta * (s.F - group.floatMean) - } - - case parser.TOPK: - // We build a heap of up to k elements, with the smallest element at heap[0]. - switch { - case int64(len(group.heap)) < k: - heap.Push(&group.heap, &Sample{ - F: s.F, - Metric: s.Metric, - }) - case group.heap[0].F < s.F || (math.IsNaN(group.heap[0].F) && !math.IsNaN(s.F)): - // This new element is bigger than the previous smallest element - overwrite that. - group.heap[0] = Sample{ - F: s.F, - Metric: s.Metric, - } - if k > 1 { - heap.Fix(&group.heap, 0) // Maintain the heap invariant. - } - } - - case parser.BOTTOMK: - // We build a heap of up to k elements, with the biggest element at heap[0]. - switch { - case int64(len(group.reverseHeap)) < k: - heap.Push(&group.reverseHeap, &Sample{ - F: s.F, - Metric: s.Metric, - }) - case group.reverseHeap[0].F > s.F || (math.IsNaN(group.reverseHeap[0].F) && !math.IsNaN(s.F)): - // This new element is smaller than the previous biggest element - overwrite that. - group.reverseHeap[0] = Sample{ - F: s.F, - Metric: s.Metric, - } - if k > 1 { - heap.Fix(&group.reverseHeap, 0) // Maintain the heap invariant. - } + group.floatValue += delta * (f - group.floatMean) } case parser.QUANTILE: - group.heap = append(group.heap, s) + group.heap = append(group.heap, Sample{F: f}) default: panic(fmt.Errorf("expected aggregation operator but got %q", op)) } } - // Construct the result Vector from the aggregated groups. - for _, aggr := range orderedResult { + // Construct the output matrix from the aggregated groups. + numSteps := int((ev.endTimestamp-ev.startTimestamp)/ev.interval) + 1 + + for ri, aggr := range groups { + if !aggr.seen { + continue + } switch op { case parser.AVG: if aggr.hasFloat && aggr.hasHistogram { @@ -2879,12 +2881,12 @@ func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, par continue } if aggr.hasHistogram { - aggr.histogramValue = aggr.histogramMean.Compact(0) + aggr.histogramValue = aggr.histogramValue.Compact(0) } else { aggr.floatValue = aggr.floatMean } - case parser.COUNT, parser.COUNT_VALUES: + case parser.COUNT: aggr.floatValue = float64(aggr.groupCount) case parser.STDVAR: @@ -2893,36 +2895,7 @@ func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, par case parser.STDDEV: aggr.floatValue = math.Sqrt(aggr.floatValue / float64(aggr.groupCount)) - case parser.TOPK: - // The heap keeps the lowest value on top, so reverse it. - if len(aggr.heap) > 1 { - sort.Sort(sort.Reverse(aggr.heap)) - } - for _, v := range aggr.heap { - enh.Out = append(enh.Out, Sample{ - Metric: v.Metric, - F: v.F, - }) - } - continue // Bypass default append. - - case parser.BOTTOMK: - // The heap keeps the highest value on top, so reverse it. - if len(aggr.reverseHeap) > 1 { - sort.Sort(sort.Reverse(aggr.reverseHeap)) - } - for _, v := range aggr.reverseHeap { - enh.Out = append(enh.Out, Sample{ - Metric: v.Metric, - F: v.F, - }) - } - continue // Bypass default append. - case parser.QUANTILE: - if math.IsNaN(q) || q < 0 || q > 1 { - annos.Add(annotations.NewInvalidQuantileWarning(q, e.Param.PositionRange())) - } aggr.floatValue = quantile(q, aggr.heap) case parser.SUM: @@ -2938,13 +2911,196 @@ func (ev *evaluator) aggregation(e *parser.AggregateExpr, grouping []string, par // For other aggregations, we already have the right value. } + ss := &outputMatrix[ri] + addToSeries(ss, enh.Ts, aggr.floatValue, aggr.histogramValue, numSteps) + } + + return annos +} + +// aggregationK evaluates topk or bottomk at one timestep on inputMatrix. +// Output that has the same labels as the input, but just k of them per group. +// seriesToResult maps inputMatrix indexes to groups indexes. +// For an instant query, returns a Matrix in descending order for topk or ascending for bottomk. +// For a range query, aggregates output in the seriess map. +func (ev *evaluator) aggregationK(e *parser.AggregateExpr, k int, inputMatrix Matrix, seriesToResult []int, groups []groupedAggregation, enh *EvalNodeHelper, seriess map[uint64]Series) (Matrix, annotations.Annotations) { + op := e.Op + var s Sample + var annos annotations.Annotations + for i := range groups { + groups[i].seen = false + } + + for si := range inputMatrix { + f, _, ok := ev.nextValues(enh.Ts, &inputMatrix[si]) + if !ok { + continue + } + s = Sample{Metric: inputMatrix[si].Metric, F: f} + + group := &groups[seriesToResult[si]] + // Initialize this group if it's the first time we've seen it. + if !group.seen { + *group = groupedAggregation{ + seen: true, + heap: make(vectorByValueHeap, 1, k), + } + group.heap[0] = s + continue + } + + switch op { + case parser.TOPK: + // We build a heap of up to k elements, with the smallest element at heap[0]. + switch { + case len(group.heap) < k: + heap.Push(&group.heap, &s) + case group.heap[0].F < s.F || (math.IsNaN(group.heap[0].F) && !math.IsNaN(s.F)): + // This new element is bigger than the previous smallest element - overwrite that. + group.heap[0] = s + if k > 1 { + heap.Fix(&group.heap, 0) // Maintain the heap invariant. + } + } + + case parser.BOTTOMK: + // We build a heap of up to k elements, with the biggest element at heap[0]. + switch { + case len(group.heap) < k: + heap.Push((*vectorByReverseValueHeap)(&group.heap), &s) + case group.heap[0].F > s.F || (math.IsNaN(group.heap[0].F) && !math.IsNaN(s.F)): + // This new element is smaller than the previous biggest element - overwrite that. + group.heap[0] = s + if k > 1 { + heap.Fix((*vectorByReverseValueHeap)(&group.heap), 0) // Maintain the heap invariant. + } + } + + default: + panic(fmt.Errorf("expected aggregation operator but got %q", op)) + } + } + + // Construct the result from the aggregated groups. + numSteps := int((ev.endTimestamp-ev.startTimestamp)/ev.interval) + 1 + var mat Matrix + if ev.endTimestamp == ev.startTimestamp { + mat = make(Matrix, 0, len(groups)) + } + + add := func(lbls labels.Labels, f float64) { + // If this could be an instant query, add directly to the matrix so the result is in consistent order. + if ev.endTimestamp == ev.startTimestamp { + mat = append(mat, Series{Metric: lbls, Floats: []FPoint{{T: enh.Ts, F: f}}}) + } else { + // Otherwise the results are added into seriess elements. + hash := lbls.Hash() + ss, ok := seriess[hash] + if !ok { + ss = Series{Metric: lbls} + } + addToSeries(&ss, enh.Ts, f, nil, numSteps) + seriess[hash] = ss + } + } + for _, aggr := range groups { + if !aggr.seen { + continue + } + switch op { + case parser.TOPK: + // The heap keeps the lowest value on top, so reverse it. + if len(aggr.heap) > 1 { + sort.Sort(sort.Reverse(aggr.heap)) + } + for _, v := range aggr.heap { + add(v.Metric, v.F) + } + + case parser.BOTTOMK: + // The heap keeps the highest value on top, so reverse it. + if len(aggr.heap) > 1 { + sort.Sort(sort.Reverse((*vectorByReverseValueHeap)(&aggr.heap))) + } + for _, v := range aggr.heap { + add(v.Metric, v.F) + } + } + } + + return mat, annos +} + +// aggregationK evaluates count_values on vec. +// Outputs as many series per group as there are values in the input. +func (ev *evaluator) aggregationCountValues(e *parser.AggregateExpr, grouping []string, valueLabel string, vec Vector, enh *EvalNodeHelper) (Vector, annotations.Annotations) { + type groupCount struct { + labels labels.Labels + count int + } + result := map[uint64]*groupCount{} + + var buf []byte + for _, s := range vec { + enh.resetBuilder(s.Metric) + enh.lb.Set(valueLabel, strconv.FormatFloat(s.F, 'f', -1, 64)) + metric := enh.lb.Labels() + + // Considering the count_values() + // operator is less frequently used than other aggregations, we're fine having to + // re-compute the grouping key on each step for this case. + var groupingKey uint64 + groupingKey, buf = generateGroupingKey(metric, grouping, e.Without, buf) + + group, ok := result[groupingKey] + // Add a new group if it doesn't exist. + if !ok { + result[groupingKey] = &groupCount{ + labels: generateGroupingLabels(enh, metric, e.Without, grouping), + count: 1, + } + continue + } + + group.count++ + } + + // Construct the result Vector from the aggregated groups. + for _, aggr := range result { enh.Out = append(enh.Out, Sample{ Metric: aggr.labels, - F: aggr.floatValue, - H: aggr.histogramValue, + F: float64(aggr.count), }) } - return enh.Out, annos + return enh.Out, nil +} + +func addToSeries(ss *Series, ts int64, f float64, h *histogram.FloatHistogram, numSteps int) { + if h == nil { + if ss.Floats == nil { + ss.Floats = getFPointSlice(numSteps) + } + ss.Floats = append(ss.Floats, FPoint{T: ts, F: f}) + return + } + if ss.Histograms == nil { + ss.Histograms = getHPointSlice(numSteps) + } + ss.Histograms = append(ss.Histograms, HPoint{T: ts, H: h}) +} + +func (ev *evaluator) nextValues(ts int64, series *Series) (f float64, h *histogram.FloatHistogram, b bool) { + switch { + case len(series.Floats) > 0 && series.Floats[0].T == ts: + f = series.Floats[0].F + series.Floats = series.Floats[1:] // Move input vectors forward + case len(series.Histograms) > 0 && series.Histograms[0].T == ts: + h = series.Histograms[0].H + series.Histograms = series.Histograms[1:] + default: + return f, h, false + } + return f, h, true } // groupingKey builds and returns the grouping key for the given metric and @@ -2962,6 +3118,21 @@ func generateGroupingKey(metric labels.Labels, grouping []string, without bool, return metric.HashForLabels(buf, grouping...) } +func generateGroupingLabels(enh *EvalNodeHelper, metric labels.Labels, without bool, grouping []string) labels.Labels { + enh.resetBuilder(metric) + switch { + case without: + enh.lb.Del(grouping...) + enh.lb.Del(labels.MetricName) + return enh.lb.Labels() + case len(grouping) > 0: + enh.lb.Keep(grouping...) + return enh.lb.Labels() + default: + return labels.EmptyLabels() + } +} + // btos returns 1 if b is true, 0 otherwise. func btos(b bool) float64 { if b { diff --git a/promql/engine_test.go b/promql/engine_test.go index ea65e01a63..0202c15ae1 100644 --- a/promql/engine_test.go +++ b/promql/engine_test.go @@ -755,6 +755,7 @@ load 10s metricWith3SampleEvery10Seconds{a="1",b="1"} 1+1x100 metricWith3SampleEvery10Seconds{a="2",b="2"} 1+1x100 metricWith3SampleEvery10Seconds{a="3",b="2"} 1+1x100 + metricWith1HistogramEvery10Seconds {{schema:1 count:5 sum:20 buckets:[1 2 1 1]}}+{{schema:1 count:10 sum:5 buckets:[1 2 3 4]}}x100 `) t.Cleanup(func() { storage.Close() }) @@ -795,6 +796,15 @@ load 10s 21000: 1, }, }, + { + Query: "metricWith1HistogramEvery10Seconds", + Start: time.Unix(21, 0), + PeakSamples: 12, + TotalSamples: 12, // 1 histogram sample of size 12 / 10 seconds + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 21000: 12, + }, + }, { // timestamp function has a special handling. Query: "timestamp(metricWith1SampleEvery10Seconds)", @@ -805,6 +815,15 @@ load 10s 21000: 1, }, }, + { + Query: "timestamp(metricWith1HistogramEvery10Seconds)", + Start: time.Unix(21, 0), + PeakSamples: 13, // histogram size 12 + 1 extra because of timestamp + TotalSamples: 1, // 1 float sample (because of timestamp) / 10 seconds + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 21000: 1, + }, + }, { Query: "metricWith1SampleEvery10Seconds", Start: time.Unix(22, 0), @@ -877,11 +896,20 @@ load 10s 201000: 6, }, }, + { + Query: "metricWith1HistogramEvery10Seconds[60s]", + Start: time.Unix(201, 0), + PeakSamples: 72, + TotalSamples: 72, // 1 histogram (size 12) / 10 seconds * 60 seconds + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 201000: 72, + }, + }, { Query: "max_over_time(metricWith1SampleEvery10Seconds[59s])[20s:5s]", Start: time.Unix(201, 0), PeakSamples: 10, - TotalSamples: 24, // (1 sample / 10 seconds * 60 seconds) * 60/5 (using 59s so we always return 6 samples + TotalSamples: 24, // (1 sample / 10 seconds * 60 seconds) * 20/5 (using 59s so we always return 6 samples // as if we run a query on 00 looking back 60 seconds we will return 7 samples; // see next test). TotalSamplesPerStep: stats.TotalSamplesPerStep{ @@ -892,12 +920,22 @@ load 10s Query: "max_over_time(metricWith1SampleEvery10Seconds[60s])[20s:5s]", Start: time.Unix(201, 0), PeakSamples: 11, - TotalSamples: 26, // (1 sample / 10 seconds * 60 seconds) + 2 as + TotalSamples: 26, // (1 sample / 10 seconds * 60 seconds) * 4 + 2 as // max_over_time(metricWith1SampleEvery10Seconds[60s]) @ 190 and 200 will return 7 samples. TotalSamplesPerStep: stats.TotalSamplesPerStep{ 201000: 26, }, }, + { + Query: "max_over_time(metricWith1HistogramEvery10Seconds[60s])[20s:5s]", + Start: time.Unix(201, 0), + PeakSamples: 72, + TotalSamples: 312, // (1 histogram (size 12) / 10 seconds * 60 seconds) * 4 + 2 * 12 as + // max_over_time(metricWith1SampleEvery10Seconds[60s]) @ 190 and 200 will return 7 samples. + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 201000: 312, + }, + }, { Query: "metricWith1SampleEvery10Seconds[60s] @ 30", Start: time.Unix(201, 0), @@ -907,6 +945,15 @@ load 10s 201000: 4, }, }, + { + Query: "metricWith1HistogramEvery10Seconds[60s] @ 30", + Start: time.Unix(201, 0), + PeakSamples: 48, + TotalSamples: 48, // @ modifier force the evaluation to at 30 seconds - So it brings 4 datapoints (0, 10, 20, 30 seconds) * 1 series + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 201000: 48, + }, + }, { Query: "sum(max_over_time(metricWith3SampleEvery10Seconds[60s] @ 30))", Start: time.Unix(201, 0), @@ -919,7 +966,7 @@ load 10s { Query: "sum by (b) (max_over_time(metricWith3SampleEvery10Seconds[60s] @ 30))", Start: time.Unix(201, 0), - PeakSamples: 8, + PeakSamples: 7, TotalSamples: 12, // @ modifier force the evaluation to at 30 seconds - So it brings 4 datapoints (0, 10, 20, 30 seconds) * 3 series TotalSamplesPerStep: stats.TotalSamplesPerStep{ 201000: 12, @@ -1035,13 +1082,42 @@ load 10s }, }, { - // timestamp function as a special handling + Query: `metricWith1HistogramEvery10Seconds`, + Start: time.Unix(204, 0), + End: time.Unix(223, 0), + Interval: 5 * time.Second, + PeakSamples: 48, + TotalSamples: 48, // 1 histogram (size 12) per query * 4 steps + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 204000: 12, // aligned to the step time, not the sample time + 209000: 12, + 214000: 12, + 219000: 12, + }, + }, + { + // timestamp function has a special handling Query: "timestamp(metricWith1SampleEvery10Seconds)", Start: time.Unix(201, 0), End: time.Unix(220, 0), Interval: 5 * time.Second, PeakSamples: 5, - TotalSamples: 4, // (1 sample / 10 seconds) * 4 steps + TotalSamples: 4, // 1 sample per query * 4 steps + TotalSamplesPerStep: stats.TotalSamplesPerStep{ + 201000: 1, + 206000: 1, + 211000: 1, + 216000: 1, + }, + }, + { + // timestamp function has a special handling + Query: "timestamp(metricWith1HistogramEvery10Seconds)", + Start: time.Unix(201, 0), + End: time.Unix(220, 0), + Interval: 5 * time.Second, + PeakSamples: 16, + TotalSamples: 4, // 1 sample per query * 4 steps TotalSamplesPerStep: stats.TotalSamplesPerStep{ 201000: 1, 206000: 1, @@ -3438,7 +3514,39 @@ func TestNativeHistogram_HistogramStdDevVar(t *testing.T) { }, NegativeBuckets: []int64{1, 0}, }, - stdVar: 1544.8582535368798, // actual variance: 1738.4082 + stdVar: 1844.4651144196398, // actual variance: 1738.4082 + }, + { + name: "-100000, -10000, -1000, -888, -888, -100, -50, -9, -8, -3", + h: &histogram.Histogram{ + Count: 10, + ZeroCount: 0, + Sum: -112946, + Schema: 0, + NegativeSpans: []histogram.Span{ + {Offset: 2, Length: 3}, + {Offset: 1, Length: 2}, + {Offset: 2, Length: 1}, + {Offset: 3, Length: 1}, + {Offset: 2, Length: 1}, + }, + NegativeBuckets: []int64{1, 0, 0, 0, 0, 2, -2, 0}, + }, + stdVar: 759352122.1939945, // actual variance: 882690990 + }, + { + name: "-10 x10", + h: &histogram.Histogram{ + Count: 10, + ZeroCount: 0, + Sum: -100, + Schema: 0, + NegativeSpans: []histogram.Span{ + {Offset: 4, Length: 1}, + }, + NegativeBuckets: []int64{10}, + }, + stdVar: 1.725830020304794, // actual variance: 0 }, { name: "-50, -8, 0, 3, 8, 9, 100, NaN", diff --git a/promql/functions.go b/promql/functions.go index d5840374e7..0cfbcc3452 100644 --- a/promql/functions.go +++ b/promql/functions.go @@ -1111,11 +1111,17 @@ func funcHistogramStdDev(vals []parser.Value, args parser.Expressions, enh *Eval it := sample.H.AllBucketIterator() for it.Next() { bucket := it.At() + if bucket.Count == 0 { + continue + } var val float64 if bucket.Lower <= 0 && 0 <= bucket.Upper { val = 0 } else { val = math.Sqrt(bucket.Upper * bucket.Lower) + if bucket.Upper < 0 { + val = -val + } } delta := val - mean variance, cVariance = kahanSumInc(bucket.Count*delta*delta, variance, cVariance) @@ -1144,11 +1150,17 @@ func funcHistogramStdVar(vals []parser.Value, args parser.Expressions, enh *Eval it := sample.H.AllBucketIterator() for it.Next() { bucket := it.At() + if bucket.Count == 0 { + continue + } var val float64 if bucket.Lower <= 0 && 0 <= bucket.Upper { val = 0 } else { val = math.Sqrt(bucket.Upper * bucket.Lower) + if bucket.Upper < 0 { + val = -val + } } delta := val - mean variance, cVariance = kahanSumInc(bucket.Count*delta*delta, variance, cVariance) diff --git a/promql/parser/parse_test.go b/promql/parser/parse_test.go index 48d7426538..c56d845947 100644 --- a/promql/parser/parse_test.go +++ b/promql/parser/parse_test.go @@ -3696,9 +3696,17 @@ func makeInt64Pointer(val int64) *int64 { return valp } +func readable(s string) string { + const maxReadableStringLen = 40 + if len(s) < maxReadableStringLen { + return s + } + return s[:maxReadableStringLen] + "..." +} + func TestParseExpressions(t *testing.T) { for _, test := range testExpr { - t.Run(test.input, func(t *testing.T) { + t.Run(readable(test.input), func(t *testing.T) { expr, err := ParseExpr(test.input) // Unexpected errors are always caused by a bug. @@ -3706,7 +3714,31 @@ func TestParseExpressions(t *testing.T) { if !test.fail { require.NoError(t, err) - require.Equal(t, test.expected, expr, "error on input '%s'", test.input) + expected := test.expected + + // The FastRegexMatcher is not comparable with a deep equal, so only compare its String() version. + if actualVector, ok := expr.(*VectorSelector); ok { + require.IsType(t, &VectorSelector{}, test.expected, "error on input '%s'", test.input) + expectedVector := test.expected.(*VectorSelector) + + require.Len(t, actualVector.LabelMatchers, len(expectedVector.LabelMatchers), "error on input '%s'", test.input) + + for i := 0; i < len(actualVector.LabelMatchers); i++ { + expectedMatcher := expectedVector.LabelMatchers[i].String() + actualMatcher := actualVector.LabelMatchers[i].String() + + require.Equal(t, expectedMatcher, actualMatcher, "unexpected label matcher '%s' on input '%s'", actualMatcher, test.input) + } + + // Make a shallow copy of the expected expr (because the test cases are defined in a global variable) + // and then reset the LabelMatcher to not compared them with the following deep equal. + expectedCopy := *expectedVector + expectedCopy.LabelMatchers = nil + expected = &expectedCopy + actualVector.LabelMatchers = nil + } + + require.Equal(t, expected, expr, "error on input '%s'", test.input) } else { require.Error(t, err) require.Contains(t, err.Error(), test.errMsg, "unexpected error on input '%s', expected '%s', got '%s'", test.input, test.errMsg, err.Error()) diff --git a/promql/test.go b/promql/test.go index 589b1e5b6b..2a9a4a2b81 100644 --- a/promql/test.go +++ b/promql/test.go @@ -46,6 +46,7 @@ var ( patSpace = regexp.MustCompile("[\t ]+") patLoad = regexp.MustCompile(`^load\s+(.+?)$`) patEvalInstant = regexp.MustCompile(`^eval(?:_(fail|ordered))?\s+instant\s+(?:at\s+(.+?))?\s+(.+)$`) + patEvalRange = regexp.MustCompile(`^eval(?:_(fail))?\s+range\s+from\s+(.+)\s+to\s+(.+)\s+step\s+(.+?)\s+(.+)$`) ) const ( @@ -72,7 +73,7 @@ func LoadedStorage(t testutil.T, input string) *teststorage.TestStorage { } // RunBuiltinTests runs an acceptance test suite against the provided engine. -func RunBuiltinTests(t *testing.T, engine engineQuerier) { +func RunBuiltinTests(t *testing.T, engine QueryEngine) { t.Cleanup(func() { parser.EnableExperimentalFunctions = false }) parser.EnableExperimentalFunctions = true @@ -89,11 +90,19 @@ func RunBuiltinTests(t *testing.T, engine engineQuerier) { } // RunTest parses and runs the test against the provided engine. -func RunTest(t testutil.T, input string, engine engineQuerier) { - test, err := newTest(t, input) - require.NoError(t, err) +func RunTest(t testutil.T, input string, engine QueryEngine) { + require.NoError(t, runTest(t, input, engine)) +} +func runTest(t testutil.T, input string, engine QueryEngine) error { + test, err := newTest(t, input) + + // Why do this before checking err? newTest() can create the test storage and then return an error, + // and we want to make sure to clean that up to avoid leaking goroutines. defer func() { + if test == nil { + return + } if test.storage != nil { test.storage.Close() } @@ -102,11 +111,19 @@ func RunTest(t testutil.T, input string, engine engineQuerier) { } }() - for _, cmd := range test.cmds { - // TODO(fabxc): aggregate command errors, yield diffs for result - // comparison errors. - require.NoError(t, test.exec(cmd, engine)) + if err != nil { + return err } + + for _, cmd := range test.cmds { + if err := test.exec(cmd, engine); err != nil { + // TODO(fabxc): aggregate command errors, yield diffs for result + // comparison errors. + return err + } + } + + return nil } // test is a sequence of read and write commands that are run @@ -137,11 +154,6 @@ func newTest(t testutil.T, input string) (*test, error) { //go:embed testdata var testsFs embed.FS -type engineQuerier interface { - NewRangeQuery(ctx context.Context, q storage.Queryable, opts QueryOpts, qs string, start, end time.Time, interval time.Duration) (Query, error) - NewInstantQuery(ctx context.Context, q storage.Queryable, opts QueryOpts, qs string, ts time.Time) (Query, error) -} - func raise(line int, format string, v ...interface{}) error { return &parser.ParseErr{ LineOffset: line, @@ -188,15 +200,26 @@ func parseSeries(defLine string, line int) (labels.Labels, []parser.SequenceValu } func (t *test) parseEval(lines []string, i int) (int, *evalCmd, error) { - if !patEvalInstant.MatchString(lines[i]) { - return i, nil, raise(i, "invalid evaluation command. (eval[_fail|_ordered] instant [at ] ") + instantParts := patEvalInstant.FindStringSubmatch(lines[i]) + rangeParts := patEvalRange.FindStringSubmatch(lines[i]) + + if instantParts == nil && rangeParts == nil { + return i, nil, raise(i, "invalid evaluation command. Must be either 'eval[_fail|_ordered] instant [at ] ' or 'eval[_fail] range from to step '") } - parts := patEvalInstant.FindStringSubmatch(lines[i]) - var ( - mod = parts[1] - at = parts[2] - expr = parts[3] - ) + + isInstant := instantParts != nil + + var mod string + var expr string + + if isInstant { + mod = instantParts[1] + expr = instantParts[3] + } else { + mod = rangeParts[1] + expr = rangeParts[5] + } + _, err := parser.ParseExpr(expr) if err != nil { parser.EnrichParseError(err, func(parseErr *parser.ParseErr) { @@ -209,15 +232,54 @@ func (t *test) parseEval(lines []string, i int) (int, *evalCmd, error) { return i, nil, err } - offset, err := model.ParseDuration(at) - if err != nil { - return i, nil, raise(i, "invalid step definition %q: %s", parts[1], err) - } - ts := testStartTime.Add(time.Duration(offset)) + formatErr := func(format string, args ...any) error { + combinedArgs := []any{expr, i + 1} + + combinedArgs = append(combinedArgs, args...) + return fmt.Errorf("error in eval %s (line %v): "+format, combinedArgs...) + } + + var cmd *evalCmd + + if isInstant { + at := instantParts[2] + offset, err := model.ParseDuration(at) + if err != nil { + return i, nil, formatErr("invalid timestamp definition %q: %s", at, err) + } + ts := testStartTime.Add(time.Duration(offset)) + cmd = newInstantEvalCmd(expr, ts, i+1) + } else { + from := rangeParts[2] + to := rangeParts[3] + step := rangeParts[4] + + parsedFrom, err := model.ParseDuration(from) + if err != nil { + return i, nil, formatErr("invalid start timestamp definition %q: %s", from, err) + } + + parsedTo, err := model.ParseDuration(to) + if err != nil { + return i, nil, formatErr("invalid end timestamp definition %q: %s", to, err) + } + + if parsedTo < parsedFrom { + return i, nil, formatErr("invalid test definition, end timestamp (%s) is before start timestamp (%s)", to, from) + } + + parsedStep, err := model.ParseDuration(step) + if err != nil { + return i, nil, formatErr("invalid step definition %q: %s", step, err) + } + + cmd = newRangeEvalCmd(expr, testStartTime.Add(time.Duration(parsedFrom)), testStartTime.Add(time.Duration(parsedTo)), time.Duration(parsedStep), i+1) + } - cmd := newEvalCmd(expr, ts, i+1) switch mod { case "ordered": + // Ordered results are not supported for range queries, but the regex for range query commands does not allow + // asserting an ordered result, so we don't need to do any error checking here. cmd.ordered = true case "fail": cmd.fail = true @@ -240,8 +302,8 @@ func (t *test) parseEval(lines []string, i int) (int, *evalCmd, error) { } // Currently, we are not expecting any matrices. - if len(vals) > 1 { - return i, nil, raise(i, "expecting multiple values in instant evaluation not allowed") + if len(vals) > 1 && isInstant { + return i, nil, formatErr("expecting multiple values in instant evaluation not allowed") } cmd.expectMetric(j, metric, vals...) } @@ -375,8 +437,11 @@ func appendSample(a storage.Appender, s Sample, m labels.Labels) error { type evalCmd struct { expr string start time.Time + end time.Time + step time.Duration line int + isRange bool // if false, instant query fail, ordered bool metrics map[uint64]labels.Labels @@ -392,7 +457,7 @@ func (e entry) String() string { return fmt.Sprintf("%d: %s", e.pos, e.vals) } -func newEvalCmd(expr string, start time.Time, line int) *evalCmd { +func newInstantEvalCmd(expr string, start time.Time, line int) *evalCmd { return &evalCmd{ expr: expr, start: start, @@ -403,6 +468,20 @@ func newEvalCmd(expr string, start time.Time, line int) *evalCmd { } } +func newRangeEvalCmd(expr string, start, end time.Time, step time.Duration, line int) *evalCmd { + return &evalCmd{ + expr: expr, + start: start, + end: end, + step: step, + line: line, + isRange: true, + + metrics: map[uint64]labels.Labels{}, + expected: map[uint64]entry{}, + } +} + func (ev *evalCmd) String() string { return "eval" } @@ -425,14 +504,88 @@ func (ev *evalCmd) expectMetric(pos int, m labels.Labels, vals ...parser.Sequenc func (ev *evalCmd) compareResult(result parser.Value) error { switch val := result.(type) { case Matrix: - return errors.New("received range result on instant evaluation") + if ev.ordered { + return fmt.Errorf("expected ordered result, but query returned a matrix") + } + + if err := assertMatrixSorted(val); err != nil { + return err + } + + seen := map[uint64]bool{} + for _, s := range val { + hash := s.Metric.Hash() + if _, ok := ev.metrics[hash]; !ok { + return fmt.Errorf("unexpected metric %s in result, has %s", s.Metric, formatSeriesResult(s)) + } + seen[hash] = true + exp := ev.expected[hash] + + var expectedFloats []FPoint + var expectedHistograms []HPoint + + for i, e := range exp.vals { + ts := ev.start.Add(time.Duration(i) * ev.step) + + if ts.After(ev.end) { + return fmt.Errorf("expected %v points for %s, but query time range cannot return this many points", len(exp.vals), ev.metrics[hash]) + } + + t := ts.UnixNano() / int64(time.Millisecond/time.Nanosecond) + + if e.Histogram != nil { + expectedHistograms = append(expectedHistograms, HPoint{T: t, H: e.Histogram}) + } else if !e.Omitted { + expectedFloats = append(expectedFloats, FPoint{T: t, F: e.Value}) + } + } + + if len(expectedFloats) != len(s.Floats) || len(expectedHistograms) != len(s.Histograms) { + return fmt.Errorf("expected %v float points and %v histogram points for %s, but got %s", len(expectedFloats), len(expectedHistograms), ev.metrics[hash], formatSeriesResult(s)) + } + + for i, expected := range expectedFloats { + actual := s.Floats[i] + + if expected.T != actual.T { + return fmt.Errorf("expected float value at index %v for %s to have timestamp %v, but it had timestamp %v (result has %s)", i, ev.metrics[hash], expected.T, actual.T, formatSeriesResult(s)) + } + + if !almostEqual(actual.F, expected.F, defaultEpsilon) { + return fmt.Errorf("expected float value at index %v (t=%v) for %s to be %v, but got %v (result has %s)", i, actual.T, ev.metrics[hash], expected.F, actual.F, formatSeriesResult(s)) + } + } + + for i, expected := range expectedHistograms { + actual := s.Histograms[i] + + if expected.T != actual.T { + return fmt.Errorf("expected histogram value at index %v for %s to have timestamp %v, but it had timestamp %v (result has %s)", i, ev.metrics[hash], expected.T, actual.T, formatSeriesResult(s)) + } + + if !actual.H.Equals(expected.H.Compact(0)) { + return fmt.Errorf("expected histogram value at index %v (t=%v) for %s to be %v, but got %v (result has %s)", i, actual.T, ev.metrics[hash], expected.H, actual.H, formatSeriesResult(s)) + } + } + + } + + for hash := range ev.expected { + if !seen[hash] { + return fmt.Errorf("expected metric %s not found", ev.metrics[hash]) + } + } case Vector: seen := map[uint64]bool{} for pos, v := range val { fp := v.Metric.Hash() if _, ok := ev.metrics[fp]; !ok { - return fmt.Errorf("unexpected metric %s in result", v.Metric) + if v.H != nil { + return fmt.Errorf("unexpected metric %s in result, has value %v", v.Metric, v.H) + } + + return fmt.Errorf("unexpected metric %s in result, has value %v", v.Metric, v.F) } exp := ev.expected[fp] if ev.ordered && exp.pos != pos+1 { @@ -440,7 +593,13 @@ func (ev *evalCmd) compareResult(result parser.Value) error { } exp0 := exp.vals[0] expH := exp0.Histogram - if (expH == nil) != (v.H == nil) || (expH != nil && !expH.Equals(v.H)) { + if expH == nil && v.H != nil { + return fmt.Errorf("expected float value %v for %s but got histogram %s", exp0, v.Metric, HistogramTestExpression(v.H)) + } + if expH != nil && v.H == nil { + return fmt.Errorf("expected histogram %s for %s but got float value %v", HistogramTestExpression(expH), v.Metric, v.F) + } + if expH != nil && !expH.Compact(0).Equals(v.H) { return fmt.Errorf("expected %v for %s but got %s", HistogramTestExpression(expH), v.Metric, HistogramTestExpression(v.H)) } if !almostEqual(exp0.Value, v.F, defaultEpsilon) { @@ -451,10 +610,6 @@ func (ev *evalCmd) compareResult(result parser.Value) error { } for fp, expVals := range ev.expected { if !seen[fp] { - fmt.Println("vector result", len(val), ev.expr) - for _, ss := range val { - fmt.Println(" ", ss.Metric, ss.T, ss.F) - } return fmt.Errorf("expected metric %s with %v not found", ev.metrics[fp], expVals) } } @@ -477,6 +632,21 @@ func (ev *evalCmd) compareResult(result parser.Value) error { return nil } +func formatSeriesResult(s Series) string { + floatPlural := "s" + histogramPlural := "s" + + if len(s.Floats) == 1 { + floatPlural = "" + } + + if len(s.Histograms) == 1 { + histogramPlural = "" + } + + return fmt.Sprintf("%v float point%s %v and %v histogram point%s %v", len(s.Floats), floatPlural, s.Floats, len(s.Histograms), histogramPlural, s.Histograms) +} + // HistogramTestExpression returns TestExpression() for the given histogram or "" if the histogram is nil. func HistogramTestExpression(h *histogram.FloatHistogram) string { if h != nil { @@ -561,7 +731,7 @@ func atModifierTestCases(exprStr string, evalTime time.Time) ([]atModifierTestCa } // exec processes a single step of the test. -func (t *test) exec(tc testCommand, engine engineQuerier) error { +func (t *test) exec(tc testCommand, engine QueryEngine) error { switch cmd := tc.(type) { case *clearCmd: t.clear() @@ -578,74 +748,7 @@ func (t *test) exec(tc testCommand, engine engineQuerier) error { } case *evalCmd: - queries, err := atModifierTestCases(cmd.expr, cmd.start) - if err != nil { - return err - } - queries = append([]atModifierTestCase{{expr: cmd.expr, evalTime: cmd.start}}, queries...) - for _, iq := range queries { - q, err := engine.NewInstantQuery(t.context, t.storage, nil, iq.expr, iq.evalTime) - if err != nil { - return err - } - defer q.Close() - res := q.Exec(t.context) - if res.Err != nil { - if cmd.fail { - continue - } - return fmt.Errorf("error evaluating query %q (line %d): %w", iq.expr, cmd.line, res.Err) - } - if res.Err == nil && cmd.fail { - return fmt.Errorf("expected error evaluating query %q (line %d) but got none", iq.expr, cmd.line) - } - err = cmd.compareResult(res.Value) - if err != nil { - return fmt.Errorf("error in %s %s (line %d): %w", cmd, iq.expr, cmd.line, err) - } - - // Check query returns same result in range mode, - // by checking against the middle step. - q, err = engine.NewRangeQuery(t.context, t.storage, nil, iq.expr, iq.evalTime.Add(-time.Minute), iq.evalTime.Add(time.Minute), time.Minute) - if err != nil { - return err - } - rangeRes := q.Exec(t.context) - if rangeRes.Err != nil { - return fmt.Errorf("error evaluating query %q (line %d) in range mode: %w", iq.expr, cmd.line, rangeRes.Err) - } - defer q.Close() - if cmd.ordered { - // Ordering isn't defined for range queries. - continue - } - mat := rangeRes.Value.(Matrix) - vec := make(Vector, 0, len(mat)) - for _, series := range mat { - // We expect either Floats or Histograms. - for _, point := range series.Floats { - if point.T == timeMilliseconds(iq.evalTime) { - vec = append(vec, Sample{Metric: series.Metric, T: point.T, F: point.F}) - break - } - } - for _, point := range series.Histograms { - if point.T == timeMilliseconds(iq.evalTime) { - vec = append(vec, Sample{Metric: series.Metric, T: point.T, H: point.H}) - break - } - } - } - if _, ok := res.Value.(Scalar); ok { - err = cmd.compareResult(Scalar{V: vec[0].F}) - } else { - err = cmd.compareResult(vec) - } - if err != nil { - return fmt.Errorf("error in %s %s (line %d) range mode: %w", cmd, iq.expr, cmd.line, err) - } - - } + return t.execEval(cmd, engine) default: panic("promql.Test.exec: unknown test command type") @@ -653,6 +756,132 @@ func (t *test) exec(tc testCommand, engine engineQuerier) error { return nil } +func (t *test) execEval(cmd *evalCmd, engine QueryEngine) error { + if cmd.isRange { + return t.execRangeEval(cmd, engine) + } + + return t.execInstantEval(cmd, engine) +} + +func (t *test) execRangeEval(cmd *evalCmd, engine QueryEngine) error { + q, err := engine.NewRangeQuery(t.context, t.storage, nil, cmd.expr, cmd.start, cmd.end, cmd.step) + if err != nil { + return fmt.Errorf("error creating range query for %q (line %d): %w", cmd.expr, cmd.line, err) + } + res := q.Exec(t.context) + if res.Err != nil { + if cmd.fail { + return nil + } + + return fmt.Errorf("error evaluating query %q (line %d): %w", cmd.expr, cmd.line, res.Err) + } + if res.Err == nil && cmd.fail { + return fmt.Errorf("expected error evaluating query %q (line %d) but got none", cmd.expr, cmd.line) + } + defer q.Close() + + if err := cmd.compareResult(res.Value); err != nil { + return fmt.Errorf("error in %s %s (line %d): %w", cmd, cmd.expr, cmd.line, err) + } + + return nil +} + +func (t *test) execInstantEval(cmd *evalCmd, engine QueryEngine) error { + queries, err := atModifierTestCases(cmd.expr, cmd.start) + if err != nil { + return err + } + queries = append([]atModifierTestCase{{expr: cmd.expr, evalTime: cmd.start}}, queries...) + for _, iq := range queries { + q, err := engine.NewInstantQuery(t.context, t.storage, nil, iq.expr, iq.evalTime) + if err != nil { + return fmt.Errorf("error creating instant query for %q (line %d): %w", cmd.expr, cmd.line, err) + } + defer q.Close() + res := q.Exec(t.context) + if res.Err != nil { + if cmd.fail { + continue + } + return fmt.Errorf("error evaluating query %q (line %d): %w", iq.expr, cmd.line, res.Err) + } + if res.Err == nil && cmd.fail { + return fmt.Errorf("expected error evaluating query %q (line %d) but got none", iq.expr, cmd.line) + } + err = cmd.compareResult(res.Value) + if err != nil { + return fmt.Errorf("error in %s %s (line %d): %w", cmd, iq.expr, cmd.line, err) + } + + // Check query returns same result in range mode, + // by checking against the middle step. + q, err = engine.NewRangeQuery(t.context, t.storage, nil, iq.expr, iq.evalTime.Add(-time.Minute), iq.evalTime.Add(time.Minute), time.Minute) + if err != nil { + return fmt.Errorf("error creating range query for %q (line %d): %w", cmd.expr, cmd.line, err) + } + rangeRes := q.Exec(t.context) + if rangeRes.Err != nil { + return fmt.Errorf("error evaluating query %q (line %d) in range mode: %w", iq.expr, cmd.line, rangeRes.Err) + } + defer q.Close() + if cmd.ordered { + // Range queries are always sorted by labels, so skip this test case that expects results in a particular order. + continue + } + mat := rangeRes.Value.(Matrix) + if err := assertMatrixSorted(mat); err != nil { + return err + } + + vec := make(Vector, 0, len(mat)) + for _, series := range mat { + // We expect either Floats or Histograms. + for _, point := range series.Floats { + if point.T == timeMilliseconds(iq.evalTime) { + vec = append(vec, Sample{Metric: series.Metric, T: point.T, F: point.F}) + break + } + } + for _, point := range series.Histograms { + if point.T == timeMilliseconds(iq.evalTime) { + vec = append(vec, Sample{Metric: series.Metric, T: point.T, H: point.H}) + break + } + } + } + if _, ok := res.Value.(Scalar); ok { + err = cmd.compareResult(Scalar{V: vec[0].F}) + } else { + err = cmd.compareResult(vec) + } + if err != nil { + return fmt.Errorf("error in %s %s (line %d) range mode: %w", cmd, iq.expr, cmd.line, err) + } + } + + return nil +} + +func assertMatrixSorted(m Matrix) error { + if len(m) <= 1 { + return nil + } + + for i, s := range m[:len(m)-1] { + nextIndex := i + 1 + nextMetric := m[nextIndex].Metric + + if labels.Compare(s.Metric, nextMetric) > 0 { + return fmt.Errorf("matrix results should always be sorted by labels, but matrix is not sorted: series at index %v with labels %s sorts before series at index %v with labels %s", nextIndex, nextMetric, i, s.Metric) + } + } + + return nil +} + // clear the current test storage of all inserted samples. func (t *test) clear() { if t.storage != nil { @@ -704,8 +933,6 @@ func parseNumber(s string) (float64, error) { // LazyLoader lazily loads samples into storage. // This is specifically implemented for unit testing of rules. type LazyLoader struct { - testutil.T - loadCmd *loadCmd storage storage.Storage @@ -727,13 +954,15 @@ type LazyLoaderOpts struct { } // NewLazyLoader returns an initialized empty LazyLoader. -func NewLazyLoader(t testutil.T, input string, opts LazyLoaderOpts) (*LazyLoader, error) { +func NewLazyLoader(input string, opts LazyLoaderOpts) (*LazyLoader, error) { ll := &LazyLoader{ - T: t, opts: opts, } err := ll.parse(input) - ll.clear() + if err != nil { + return nil, err + } + err = ll.clear() return ll, err } @@ -761,15 +990,20 @@ func (ll *LazyLoader) parse(input string) error { } // clear the current test storage of all inserted samples. -func (ll *LazyLoader) clear() { +func (ll *LazyLoader) clear() error { if ll.storage != nil { - err := ll.storage.Close() - require.NoError(ll.T, err, "Unexpected error while closing test storage.") + if err := ll.storage.Close(); err != nil { + return fmt.Errorf("closing test storage: %w", err) + } } if ll.cancelCtx != nil { ll.cancelCtx() } - ll.storage = teststorage.New(ll) + var err error + ll.storage, err = teststorage.NewWithError() + if err != nil { + return err + } opts := EngineOpts{ Logger: nil, @@ -783,6 +1017,7 @@ func (ll *LazyLoader) clear() { ll.queryEngine = NewEngine(opts) ll.context, ll.cancelCtx = context.WithCancel(context.Background()) + return nil } // appendTill appends the defined time series to the storage till the given timestamp (in milliseconds). @@ -836,8 +1071,7 @@ func (ll *LazyLoader) Storage() storage.Storage { } // Close closes resources associated with the LazyLoader. -func (ll *LazyLoader) Close() { +func (ll *LazyLoader) Close() error { ll.cancelCtx() - err := ll.storage.Close() - require.NoError(ll.T, err, "Unexpected error while closing test storage.") + return ll.storage.Close() } diff --git a/promql/test_test.go b/promql/test_test.go index ee2a0e264b..a5b24ac698 100644 --- a/promql/test_test.go +++ b/promql/test_test.go @@ -110,7 +110,7 @@ func TestLazyLoader_WithSamplesTill(t *testing.T) { } for _, c := range cases { - suite, err := NewLazyLoader(t, c.loadString, LazyLoaderOpts{}) + suite, err := NewLazyLoader(c.loadString, LazyLoaderOpts{}) require.NoError(t, err) defer suite.Close() @@ -156,3 +156,363 @@ func TestLazyLoader_WithSamplesTill(t *testing.T) { } } } + +func TestRunTest(t *testing.T) { + testData := ` +load 5m + http_requests{job="api-server", instance="0", group="production"} 0+10x10 + http_requests{job="api-server", instance="1", group="production"} 0+20x10 + http_requests{job="api-server", instance="0", group="canary"} 0+30x10 + http_requests{job="api-server", instance="1", group="canary"} 0+40x10 +` + + testCases := map[string]struct { + input string + expectedError string + }{ + "instant query with expected float result": { + input: testData + ` +eval instant at 5m sum by (group) (http_requests) + {group="production"} 30 + {group="canary"} 70 +`, + }, + "instant query with unexpected float result": { + input: testData + ` +eval instant at 5m sum by (group) (http_requests) + {group="production"} 30 + {group="canary"} 80 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected 80 for {group="canary"} but got 70`, + }, + "instant query with expected histogram result": { + input: ` +load 5m + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} + +eval instant at 0 testmetric + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} +`, + }, + "instant query with unexpected histogram result": { + input: ` +load 5m + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} + +eval instant at 0 testmetric + testmetric {{schema:-1 sum:6 count:1 buckets:[1] offset:1}} +`, + expectedError: `error in eval testmetric (line 5): expected {{schema:-1 count:1 sum:6 offset:1 buckets:[1]}} for {__name__="testmetric"} but got {{schema:-1 count:1 sum:4 offset:1 buckets:[1]}}`, + }, + "instant query with float value returned when histogram expected": { + input: ` +load 5m + testmetric 2 + +eval instant at 0 testmetric + testmetric {{}} +`, + expectedError: `error in eval testmetric (line 5): expected histogram {{}} for {__name__="testmetric"} but got float value 2`, + }, + "instant query with histogram returned when float expected": { + input: ` +load 5m + testmetric {{}} + +eval instant at 0 testmetric + testmetric 2 +`, + expectedError: `error in eval testmetric (line 5): expected float value 2.000000 for {__name__="testmetric"} but got histogram {{}}`, + }, + "instant query, but result has an unexpected series with a float value": { + input: testData + ` +eval instant at 5m sum by (group) (http_requests) + {group="production"} 30 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): unexpected metric {group="canary"} in result, has value 70`, + }, + "instant query, but result has an unexpected series with a histogram value": { + input: ` +load 5m + testmetric {{}} + +eval instant at 5m testmetric +`, + expectedError: `error in eval testmetric (line 5): unexpected metric {__name__="testmetric"} in result, has value {count:0, sum:0}`, + }, + "instant query, but result is missing a series": { + input: testData + ` +eval instant at 5m sum by (group) (http_requests) + {group="production"} 30 + {group="canary"} 70 + {group="test"} 100 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected metric {group="test"} with 3: [100.000000] not found`, + }, + "instant query expected to fail, and query fails": { + input: ` +load 5m + testmetric1{src="a",dst="b"} 0 + testmetric2{src="a",dst="b"} 1 + +eval_fail instant at 0m ceil({__name__=~'testmetric1|testmetric2'}) +`, + }, + "instant query expected to fail, but query succeeds": { + input: `eval_fail instant at 0s vector(0)`, + expectedError: `expected error evaluating query "vector(0)" (line 1) but got none`, + }, + "instant query with results expected to match provided order, and result is in expected order": { + input: testData + ` +eval_ordered instant at 50m sort(http_requests) + http_requests{group="production", instance="0", job="api-server"} 100 + http_requests{group="production", instance="1", job="api-server"} 200 + http_requests{group="canary", instance="0", job="api-server"} 300 + http_requests{group="canary", instance="1", job="api-server"} 400 +`, + }, + "instant query with results expected to match provided order, but result is out of order": { + input: testData + ` +eval_ordered instant at 50m sort(http_requests) + http_requests{group="production", instance="0", job="api-server"} 100 + http_requests{group="production", instance="1", job="api-server"} 200 + http_requests{group="canary", instance="1", job="api-server"} 400 + http_requests{group="canary", instance="0", job="api-server"} 300 +`, + expectedError: `error in eval sort(http_requests) (line 8): expected metric {__name__="http_requests", group="canary", instance="0", job="api-server"} with [300.000000] at position 4 but was at 3`, + }, + "instant query with results expected to match provided order, but result has an unexpected series": { + input: testData + ` +eval_ordered instant at 50m sort(http_requests) + http_requests{group="production", instance="0", job="api-server"} 100 + http_requests{group="production", instance="1", job="api-server"} 200 + http_requests{group="canary", instance="0", job="api-server"} 300 +`, + expectedError: `error in eval sort(http_requests) (line 8): unexpected metric {__name__="http_requests", group="canary", instance="1", job="api-server"} in result, has value 400`, + }, + "instant query with invalid timestamp": { + input: `eval instant at abc123 vector(0)`, + expectedError: `error in eval vector(0) (line 1): invalid timestamp definition "abc123": not a valid duration string: "abc123"`, + }, + "range query with expected result": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 60 + {group="canary"} 0 70 140 +`, + }, + "range query with unexpected float value": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 60 + {group="canary"} 0 80 140 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected float value at index 1 (t=300000) for {group="canary"} to be 80, but got 70 (result has 3 float points [0 @[0] 70 @[300000] 140 @[600000]] and 0 histogram points [])`, + }, + "range query with expected histogram values": { + input: ` +load 5m + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} {{schema:-1 sum:5 count:1 buckets:[1] offset:1}} {{schema:-1 sum:6 count:1 buckets:[1] offset:1}} + +eval range from 0 to 10m step 5m testmetric + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} {{schema:-1 sum:5 count:1 buckets:[1] offset:1}} {{schema:-1 sum:6 count:1 buckets:[1] offset:1}} +`, + }, + "range query with unexpected histogram value": { + input: ` +load 5m + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} {{schema:-1 sum:5 count:1 buckets:[1] offset:1}} {{schema:-1 sum:6 count:1 buckets:[1] offset:1}} + +eval range from 0 to 10m step 5m testmetric + testmetric {{schema:-1 sum:4 count:1 buckets:[1] offset:1}} {{schema:-1 sum:7 count:1 buckets:[1] offset:1}} {{schema:-1 sum:8 count:1 buckets:[1] offset:1}} +`, + expectedError: `error in eval testmetric (line 5): expected histogram value at index 1 (t=300000) for {__name__="testmetric"} to be {count:1, sum:7, (1,4]:1}, but got {count:1, sum:5, (1,4]:1} (result has 0 float points [] and 3 histogram points [{count:1, sum:4, (1,4]:1} @[0] {count:1, sum:5, (1,4]:1} @[300000] {count:1, sum:6, (1,4]:1} @[600000]])`, + }, + "range query with too many points for query time range": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 60 90 + {group="canary"} 0 70 140 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected 4 points for {group="production"}, but query time range cannot return this many points`, + }, + "range query with missing point in result": { + input: ` +load 5m + testmetric 5 + +eval range from 0 to 6m step 6m testmetric + testmetric 5 10 +`, + expectedError: `error in eval testmetric (line 5): expected 2 float points and 0 histogram points for {__name__="testmetric"}, but got 1 float point [5 @[0]] and 0 histogram points []`, + }, + "range query with extra point in result": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 + {group="canary"} 0 70 140 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected 2 float points and 0 histogram points for {group="production"}, but got 3 float points [0 @[0] 30 @[300000] 60 @[600000]] and 0 histogram points []`, + }, + "range query, but result has an unexpected series": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 60 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): unexpected metric {group="canary"} in result, has 3 float points [0 @[0] 70 @[300000] 140 @[600000]] and 0 histogram points []`, + }, + "range query, but result is missing a series": { + input: testData + ` +eval range from 0 to 10m step 5m sum by (group) (http_requests) + {group="production"} 0 30 60 + {group="canary"} 0 70 140 + {group="test"} 0 100 200 +`, + expectedError: `error in eval sum by (group) (http_requests) (line 8): expected metric {group="test"} not found`, + }, + "range query expected to fail, and query fails": { + input: ` +load 5m + testmetric1{src="a",dst="b"} 0 + testmetric2{src="a",dst="b"} 1 + +eval_fail range from 0 to 10m step 5m ceil({__name__=~'testmetric1|testmetric2'}) +`, + }, + "range query expected to fail, but query succeeds": { + input: `eval_fail range from 0 to 10m step 5m vector(0)`, + expectedError: `expected error evaluating query "vector(0)" (line 1) but got none`, + }, + "range query with from and to timestamps in wrong order": { + input: `eval range from 10m to 9m step 5m vector(0)`, + expectedError: `error in eval vector(0) (line 1): invalid test definition, end timestamp (9m) is before start timestamp (10m)`, + }, + "range query with sparse output": { + input: ` +load 6m + testmetric 1 _ 3 + +eval range from 0 to 18m step 6m testmetric + testmetric 1 _ 3 +`, + }, + "range query with float value returned when no value expected": { + input: ` +load 6m + testmetric 1 2 3 + +eval range from 0 to 18m step 6m testmetric + testmetric 1 _ 3 +`, + expectedError: `error in eval testmetric (line 5): expected 2 float points and 0 histogram points for {__name__="testmetric"}, but got 3 float points [1 @[0] 2 @[360000] 3 @[720000]] and 0 histogram points []`, + }, + "range query with float value returned when histogram expected": { + input: ` +load 5m + testmetric 2 3 + +eval range from 0 to 5m step 5m testmetric + testmetric {{}} {{}} +`, + expectedError: `error in eval testmetric (line 5): expected 0 float points and 2 histogram points for {__name__="testmetric"}, but got 2 float points [2 @[0] 3 @[300000]] and 0 histogram points []`, + }, + "range query with histogram returned when float expected": { + input: ` +load 5m + testmetric {{}} {{}} + +eval range from 0 to 5m step 5m testmetric + testmetric 2 3 +`, + expectedError: `error in eval testmetric (line 5): expected 2 float points and 0 histogram points for {__name__="testmetric"}, but got 0 float points [] and 2 histogram points [{count:0, sum:0} @[0] {count:0, sum:0} @[300000]]`, + }, + "range query with expected mixed results": { + input: ` +load 6m + testmetric{group="a"} {{}} _ _ + testmetric{group="b"} _ _ 3 + +eval range from 0 to 12m step 6m sum(testmetric) + {} {{}} _ 3 +`, + }, + "range query with mixed results and incorrect values": { + input: ` +load 5m + testmetric 3 {{}} + +eval range from 0 to 5m step 5m testmetric + testmetric {{}} 3 +`, + expectedError: `error in eval testmetric (line 5): expected float value at index 0 for {__name__="testmetric"} to have timestamp 300000, but it had timestamp 0 (result has 1 float point [3 @[0]] and 1 histogram point [{count:0, sum:0} @[300000]])`, + }, + } + + for name, testCase := range testCases { + t.Run(name, func(t *testing.T) { + err := runTest(t, testCase.input, newTestEngine()) + + if testCase.expectedError == "" { + require.NoError(t, err) + } else { + require.EqualError(t, err, testCase.expectedError) + } + }) + } +} + +func TestAssertMatrixSorted(t *testing.T) { + testCases := map[string]struct { + matrix Matrix + expectedError string + }{ + "empty matrix": { + matrix: Matrix{}, + }, + "matrix with one series": { + matrix: Matrix{ + Series{Metric: labels.FromStrings("the_label", "value_1")}, + }, + }, + "matrix with two series, series in sorted order": { + matrix: Matrix{ + Series{Metric: labels.FromStrings("the_label", "value_1")}, + Series{Metric: labels.FromStrings("the_label", "value_2")}, + }, + }, + "matrix with two series, series in reverse order": { + matrix: Matrix{ + Series{Metric: labels.FromStrings("the_label", "value_2")}, + Series{Metric: labels.FromStrings("the_label", "value_1")}, + }, + expectedError: `matrix results should always be sorted by labels, but matrix is not sorted: series at index 1 with labels {the_label="value_1"} sorts before series at index 0 with labels {the_label="value_2"}`, + }, + "matrix with three series, series in sorted order": { + matrix: Matrix{ + Series{Metric: labels.FromStrings("the_label", "value_1")}, + Series{Metric: labels.FromStrings("the_label", "value_2")}, + Series{Metric: labels.FromStrings("the_label", "value_3")}, + }, + }, + "matrix with three series, series not in sorted order": { + matrix: Matrix{ + Series{Metric: labels.FromStrings("the_label", "value_1")}, + Series{Metric: labels.FromStrings("the_label", "value_3")}, + Series{Metric: labels.FromStrings("the_label", "value_2")}, + }, + expectedError: `matrix results should always be sorted by labels, but matrix is not sorted: series at index 2 with labels {the_label="value_2"} sorts before series at index 1 with labels {the_label="value_3"}`, + }, + } + + for name, testCase := range testCases { + t.Run(name, func(t *testing.T) { + err := assertMatrixSorted(testCase.matrix) + + if testCase.expectedError == "" { + require.NoError(t, err) + } else { + require.EqualError(t, err, testCase.expectedError) + } + }) + } +} diff --git a/rules/group.go b/rules/group.go index b8694a255e..c268d2df7d 100644 --- a/rules/group.go +++ b/rules/group.go @@ -546,13 +546,13 @@ func (g *Group) Eval(ctx context.Context, ts time.Time) { } } if numOutOfOrder > 0 { - level.Warn(logger).Log("msg", "Error on ingesting out-of-order result from rule evaluation", "numDropped", numOutOfOrder) + level.Warn(logger).Log("msg", "Error on ingesting out-of-order result from rule evaluation", "num_dropped", numOutOfOrder) } if numTooOld > 0 { - level.Warn(logger).Log("msg", "Error on ingesting too old result from rule evaluation", "numDropped", numTooOld) + level.Warn(logger).Log("msg", "Error on ingesting too old result from rule evaluation", "num_dropped", numTooOld) } if numDuplicates > 0 { - level.Warn(logger).Log("msg", "Error on ingesting results from rule evaluation with different value but same timestamp", "numDropped", numDuplicates) + level.Warn(logger).Log("msg", "Error on ingesting results from rule evaluation with different value but same timestamp", "num_dropped", numDuplicates) } for metric, lset := range g.seriesInPreviousEval[i] { diff --git a/rules/manager.go b/rules/manager.go index e87d55b1e1..165dca144e 100644 --- a/rules/manager.go +++ b/rules/manager.go @@ -43,7 +43,7 @@ type QueryFunc func(ctx context.Context, q string, t time.Time) (promql.Vector, // EngineQueryFunc returns a new query function that executes instant queries against // the given engine. // It converts scalar into vector results. -func EngineQueryFunc(engine *promql.Engine, q storage.Queryable) QueryFunc { +func EngineQueryFunc(engine promql.QueryEngine, q storage.Queryable) QueryFunc { return func(ctx context.Context, qs string, t time.Time) (promql.Vector, error) { q, err := engine.NewInstantQuery(ctx, q, nil, qs, t) if err != nil { diff --git a/scrape/manager.go b/scrape/manager.go index 3ad315a50a..e805fd43af 100644 --- a/scrape/manager.go +++ b/scrape/manager.go @@ -129,6 +129,11 @@ func (m *Manager) Run(tsets <-chan map[string][]*targetgroup.Group) error { } } +// UnregisterMetrics unregisters manager metrics. +func (m *Manager) UnregisterMetrics() { + m.metrics.Unregister() +} + func (m *Manager) reloader() { reloadIntervalDuration := m.opts.DiscoveryReloadInterval if reloadIntervalDuration < model.Duration(5*time.Second) { diff --git a/scrape/manager_test.go b/scrape/manager_test.go index f90fd0ce66..51af45f8cf 100644 --- a/scrape/manager_test.go +++ b/scrape/manager_test.go @@ -857,3 +857,16 @@ func getResultFloats(app *collectResultAppender, expectedMetricName string) (res } return result } + +func TestUnregisterMetrics(t *testing.T) { + reg := prometheus.NewRegistry() + // Check that all metrics can be unregistered, allowing a second manager to be created. + for i := 0; i < 2; i++ { + opts := Options{} + manager, err := NewManager(&opts, nil, nil, reg) + require.NotNil(t, manager) + require.NoError(t, err) + // Unregister all metrics. + manager.UnregisterMetrics() + } +} diff --git a/scrape/metrics.go b/scrape/metrics.go index 7082bc743b..b67d0686b6 100644 --- a/scrape/metrics.go +++ b/scrape/metrics.go @@ -20,6 +20,7 @@ import ( ) type scrapeMetrics struct { + reg prometheus.Registerer // Used by Manager. targetMetadataCache *MetadataMetricsCollector targetScrapePools prometheus.Counter @@ -54,7 +55,7 @@ type scrapeMetrics struct { } func newScrapeMetrics(reg prometheus.Registerer) (*scrapeMetrics, error) { - sm := &scrapeMetrics{} + sm := &scrapeMetrics{reg: reg} // Manager metrics. sm.targetMetadataCache = &MetadataMetricsCollector{ @@ -260,6 +261,32 @@ func (sm *scrapeMetrics) setTargetMetadataCacheGatherer(gatherer TargetsGatherer sm.targetMetadataCache.TargetsGatherer = gatherer } +// Unregister unregisters all metrics. +func (sm *scrapeMetrics) Unregister() { + sm.reg.Unregister(sm.targetMetadataCache) + sm.reg.Unregister(sm.targetScrapePools) + sm.reg.Unregister(sm.targetScrapePoolsFailed) + sm.reg.Unregister(sm.targetReloadIntervalLength) + sm.reg.Unregister(sm.targetScrapePoolReloads) + sm.reg.Unregister(sm.targetScrapePoolReloadsFailed) + sm.reg.Unregister(sm.targetSyncIntervalLength) + sm.reg.Unregister(sm.targetScrapePoolSyncsCounter) + sm.reg.Unregister(sm.targetScrapePoolExceededTargetLimit) + sm.reg.Unregister(sm.targetScrapePoolTargetLimit) + sm.reg.Unregister(sm.targetScrapePoolTargetsAdded) + sm.reg.Unregister(sm.targetSyncFailed) + sm.reg.Unregister(sm.targetScrapeExceededBodySizeLimit) + sm.reg.Unregister(sm.targetScrapeCacheFlushForced) + sm.reg.Unregister(sm.targetIntervalLength) + sm.reg.Unregister(sm.targetScrapeSampleLimit) + sm.reg.Unregister(sm.targetScrapeSampleDuplicate) + sm.reg.Unregister(sm.targetScrapeSampleOutOfOrder) + sm.reg.Unregister(sm.targetScrapeSampleOutOfBounds) + sm.reg.Unregister(sm.targetScrapeExemplarOutOfOrder) + sm.reg.Unregister(sm.targetScrapePoolExceededLabelLimits) + sm.reg.Unregister(sm.targetScrapeNativeHistogramBucketLimit) +} + type TargetsGatherer interface { TargetsActive() map[string][]*Target } diff --git a/scrape/scrape.go b/scrape/scrape.go index 219eba4d99..4bbeab57a7 100644 --- a/scrape/scrape.go +++ b/scrape/scrape.go @@ -726,7 +726,7 @@ var UserAgent = fmt.Sprintf("Prometheus/%s", version.Version) func (s *targetScraper) scrape(ctx context.Context) (*http.Response, error) { if s.req == nil { - req, err := http.NewRequest("GET", s.URL().String(), nil) + req, err := http.NewRequest(http.MethodGet, s.URL().String(), nil) if err != nil { return nil, err } @@ -956,13 +956,14 @@ func (c *scrapeCache) iterDone(flushCache bool) { } } -func (c *scrapeCache) get(met []byte) (*cacheEntry, bool) { +func (c *scrapeCache) get(met []byte) (*cacheEntry, bool, bool) { e, ok := c.series[string(met)] if !ok { - return nil, false + return nil, false, false } + alreadyScraped := e.lastIter == c.iter e.lastIter = c.iter - return e, true + return e, true, alreadyScraped } func (c *scrapeCache) addRef(met []byte, ref storage.SeriesRef, lset labels.Labels, hash uint64) { @@ -1568,7 +1569,7 @@ loop: if sl.cache.getDropped(met) { continue } - ce, ok := sl.cache.get(met) + ce, ok, seriesAlreadyScraped := sl.cache.get(met) var ( ref storage.SeriesRef hash uint64 @@ -1577,6 +1578,7 @@ loop: if ok { ref = ce.ref lset = ce.lset + hash = ce.hash // Update metadata only if it changed in the current iteration. updateMetadata(lset, false) @@ -1613,25 +1615,36 @@ loop: updateMetadata(lset, true) } - if ctMs := p.CreatedTimestamp(); sl.enableCTZeroIngestion && ctMs != nil { - ref, err = app.AppendCTZeroSample(ref, lset, t, *ctMs) - if err != nil && !errors.Is(err, storage.ErrOutOfOrderCT) { // OOO is a common case, ignoring completely for now. - // CT is an experimental feature. For now, we don't need to fail the - // scrape on errors updating the created timestamp, log debug. - level.Debug(sl.l).Log("msg", "Error when appending CT in scrape loop", "series", string(met), "ct", *ctMs, "t", t, "err", err) + if seriesAlreadyScraped { + err = storage.ErrDuplicateSampleForTimestamp + } else { + if ctMs := p.CreatedTimestamp(); sl.enableCTZeroIngestion && ctMs != nil { + ref, err = app.AppendCTZeroSample(ref, lset, t, *ctMs) + if err != nil && !errors.Is(err, storage.ErrOutOfOrderCT) { // OOO is a common case, ignoring completely for now. + // CT is an experimental feature. For now, we don't need to fail the + // scrape on errors updating the created timestamp, log debug. + level.Debug(sl.l).Log("msg", "Error when appending CT in scrape loop", "series", string(met), "ct", *ctMs, "t", t, "err", err) + } + } + + if isHistogram { + if h != nil { + ref, err = app.AppendHistogram(ref, lset, t, h, nil) + } else { + ref, err = app.AppendHistogram(ref, lset, t, nil, fh) + } + } else { + ref, err = app.Append(ref, lset, t, val) } } - if isHistogram { - if h != nil { - ref, err = app.AppendHistogram(ref, lset, t, h, nil) - } else { - ref, err = app.AppendHistogram(ref, lset, t, nil, fh) + if err == nil { + if (parsedTimestamp == nil || sl.trackTimestampsStaleness) && ce != nil { + sl.cache.trackStaleness(ce.hash, ce.lset) } - } else { - ref, err = app.Append(ref, lset, t, val) } - sampleAdded, err = sl.checkAddError(ce, met, parsedTimestamp, err, &sampleLimitErr, &bucketLimitErr, &appErrs) + + sampleAdded, err = sl.checkAddError(met, err, &sampleLimitErr, &bucketLimitErr, &appErrs) if err != nil { if !errors.Is(err, storage.ErrNotFound) { level.Debug(sl.l).Log("msg", "Unexpected error", "series", string(met), "err", err) @@ -1652,6 +1665,8 @@ loop: // Increment added even if there's an error so we correctly report the // number of samples remaining after relabeling. + // We still report duplicated samples here since this number should be the exact number + // of time series exposed on a scrape after relabelling. added++ exemplars = exemplars[:0] // Reset and reuse the exemplar slice. for hasExemplar := p.Exemplar(&e); hasExemplar; hasExemplar = p.Exemplar(&e) { @@ -1746,12 +1761,9 @@ loop: // Adds samples to the appender, checking the error, and then returns the # of samples added, // whether the caller should continue to process more samples, and any sample or bucket limit errors. -func (sl *scrapeLoop) checkAddError(ce *cacheEntry, met []byte, tp *int64, err error, sampleLimitErr, bucketLimitErr *error, appErrs *appendErrors) (bool, error) { +func (sl *scrapeLoop) checkAddError(met []byte, err error, sampleLimitErr, bucketLimitErr *error, appErrs *appendErrors) (bool, error) { switch { case err == nil: - if (tp == nil || sl.trackTimestampsStaleness) && ce != nil { - sl.cache.trackStaleness(ce.hash, ce.lset) - } return true, nil case errors.Is(err, storage.ErrNotFound): return false, storage.ErrNotFound @@ -1874,7 +1886,7 @@ func (sl *scrapeLoop) reportStale(app storage.Appender, start time.Time) (err er } func (sl *scrapeLoop) addReportSample(app storage.Appender, s []byte, t int64, v float64, b *labels.Builder) error { - ce, ok := sl.cache.get(s) + ce, ok, _ := sl.cache.get(s) var ref storage.SeriesRef var lset labels.Labels if ok { diff --git a/scrape/scrape_test.go b/scrape/scrape_test.go index 4dfa4f684e..20b21936b9 100644 --- a/scrape/scrape_test.go +++ b/scrape/scrape_test.go @@ -1069,6 +1069,7 @@ func makeTestMetrics(n int) []byte { fmt.Fprintf(&sb, "# HELP metric_a help text\n") fmt.Fprintf(&sb, "metric_a{foo=\"%d\",bar=\"%d\"} 1\n", i, i*100) } + fmt.Fprintf(&sb, "# EOF\n") return sb.Bytes() } @@ -2636,6 +2637,9 @@ func TestScrapeLoopDiscardDuplicateLabels(t *testing.T) { _, _, _, err := sl.append(slApp, []byte("test_metric{le=\"500\"} 1\ntest_metric{le=\"600\",le=\"700\"} 1\n"), "", time.Time{}) require.Error(t, err) require.NoError(t, slApp.Rollback()) + // We need to cycle staleness cache maps after a manual rollback. Otherwise they will have old entries in them, + // which would cause ErrDuplicateSampleForTimestamp errors on the next append. + sl.cache.iterDone(true) q, err := s.Querier(time.Time{}.UnixNano(), 0) require.NoError(t, err) @@ -2972,7 +2976,7 @@ func TestReuseCacheRace(t *testing.T) { func TestCheckAddError(t *testing.T) { var appErrs appendErrors sl := scrapeLoop{l: log.NewNopLogger(), metrics: newTestScrapeMetrics(t)} - sl.checkAddError(nil, nil, nil, storage.ErrOutOfOrderSample, nil, nil, &appErrs) + sl.checkAddError(nil, storage.ErrOutOfOrderSample, nil, nil, &appErrs) require.Equal(t, 1, appErrs.numOutOfOrder) } @@ -3601,6 +3605,34 @@ func BenchmarkTargetScraperGzip(b *testing.B) { } } +// When a scrape contains multiple instances for the same time series we should increment +// prometheus_target_scrapes_sample_duplicate_timestamp_total metric. +func TestScrapeLoopSeriesAddedDuplicates(t *testing.T) { + ctx, sl := simpleTestScrapeLoop(t) + + slApp := sl.appender(ctx) + total, added, seriesAdded, err := sl.append(slApp, []byte("test_metric 1\ntest_metric 2\ntest_metric 3\n"), "", time.Time{}) + require.NoError(t, err) + require.NoError(t, slApp.Commit()) + require.Equal(t, 3, total) + require.Equal(t, 3, added) + require.Equal(t, 1, seriesAdded) + + slApp = sl.appender(ctx) + total, added, seriesAdded, err = sl.append(slApp, []byte("test_metric 1\ntest_metric 1\ntest_metric 1\n"), "", time.Time{}) + require.NoError(t, err) + require.NoError(t, slApp.Commit()) + require.Equal(t, 3, total) + require.Equal(t, 3, added) + require.Equal(t, 0, seriesAdded) + + metric := dto.Metric{} + err = sl.metrics.targetScrapeSampleDuplicate.Write(&metric) + require.NoError(t, err) + value := metric.GetCounter().GetValue() + require.Equal(t, 4.0, value) +} + // This tests running a full scrape loop and checking that the scrape option // `native_histogram_min_bucket_factor` is used correctly. func TestNativeHistogramMaxSchemaSet(t *testing.T) { diff --git a/scripts/golangci-lint.yml b/scripts/golangci-lint.yml index 4dc7b830f6..a7a40c1be5 100644 --- a/scripts/golangci-lint.yml +++ b/scripts/golangci-lint.yml @@ -24,7 +24,7 @@ jobs: runs-on: ubuntu-latest steps: - name: Checkout repository - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1 + uses: actions/checkout@9bb56186c3b09b4f86b1c65136769dd318469633 # v4.1.2 - name: install Go uses: actions/setup-go@0c52d547c9bc32b1aa3301fd7a9cb496313a4491 # v5.0.0 with: @@ -35,4 +35,4 @@ jobs: - name: Lint uses: golangci/golangci-lint-action@3cfe3a4abbb849e10058ce4af15d205b6da42804 # v4.0.0 with: - version: v1.55.2 + version: v1.56.2 diff --git a/scripts/sync_repo_files.sh b/scripts/sync_repo_files.sh index 6965a45452..6459fb1e7a 100755 --- a/scripts/sync_repo_files.sh +++ b/scripts/sync_repo_files.sh @@ -37,7 +37,7 @@ if [ -z "${GITHUB_TOKEN}" ]; then fi # List of files that should be synced. -SYNC_FILES="CODE_OF_CONDUCT.md LICENSE Makefile.common SECURITY.md .yamllint scripts/golangci-lint.yml .github/workflows/scorecards.yml" +SYNC_FILES="CODE_OF_CONDUCT.md LICENSE Makefile.common SECURITY.md .yamllint scripts/golangci-lint.yml .github/workflows/scorecards.yml .github/workflows/container_description.yml" # Go to the root of the repo cd "$(git rev-parse --show-cdup)" || exit 1 @@ -99,6 +99,15 @@ check_go() { curl -sLf -o /dev/null "https://raw.githubusercontent.com/${org_repo}/${default_branch}/go.mod" } +check_docker() { + local org_repo + local default_branch + org_repo="$1" + default_branch="$2" + + curl -sLf -o /dev/null "https://raw.githubusercontent.com/${org_repo}/${default_branch}/Dockerfile" +} + process_repo() { local org_repo local default_branch @@ -119,6 +128,10 @@ process_repo() { echo "${org_repo} is not Go, skipping golangci-lint.yml." continue fi + if [[ "${source_file}" == '.github/workflows/container_description.yml' ]] && ! check_docker "${org_repo}" "${default_branch}" ; then + echo "${org_repo} has no Dockerfile, skipping container_description.yml." + continue + fi if [[ "${source_file}" == 'LICENSE' ]] && ! check_license "${target_file}" ; then echo "LICENSE in ${org_repo} is not apache, skipping." continue @@ -131,7 +144,7 @@ process_repo() { if [[ -z "${target_file}" ]]; then echo "${target_filename} doesn't exist in ${org_repo}" case "${source_file}" in - CODE_OF_CONDUCT.md | SECURITY.md) + CODE_OF_CONDUCT.md | SECURITY.md | .github/workflows/container_description.yml) echo "${source_file} missing in ${org_repo}, force updating." needs_update+=("${source_file}") ;; diff --git a/storage/merge_test.go b/storage/merge_test.go index 05e1c75278..c3a4725aa0 100644 --- a/storage/merge_test.go +++ b/storage/merge_test.go @@ -357,12 +357,12 @@ func TestMergeChunkQuerierWithNoVerticalChunkSeriesMerger(t *testing.T) { t.Run(tc.name, func(t *testing.T) { var p ChunkQuerier if tc.primaryChkQuerierSeries != nil { - p = &mockChunkQurier{toReturn: tc.primaryChkQuerierSeries} + p = &mockChunkQuerier{toReturn: tc.primaryChkQuerierSeries} } var qs []ChunkQuerier for _, in := range tc.chkQuerierSeries { - qs = append(qs, &mockChunkQurier{toReturn: in}) + qs = append(qs, &mockChunkQuerier{toReturn: in}) } qs = append(qs, tc.extraQueriers...) @@ -934,7 +934,7 @@ func (m *mockQuerier) Select(_ context.Context, sortSeries bool, _ *SelectHints, return NewMockSeriesSet(cpy...) } -type mockChunkQurier struct { +type mockChunkQuerier struct { LabelQuerier toReturn []ChunkSeries @@ -948,7 +948,7 @@ func (a chunkSeriesByLabel) Less(i, j int) bool { return labels.Compare(a[i].Labels(), a[j].Labels()) < 0 } -func (m *mockChunkQurier) Select(_ context.Context, sortSeries bool, _ *SelectHints, _ ...*labels.Matcher) ChunkSeriesSet { +func (m *mockChunkQuerier) Select(_ context.Context, sortSeries bool, _ *SelectHints, _ ...*labels.Matcher) ChunkSeriesSet { cpy := make([]ChunkSeries, len(m.toReturn)) copy(cpy, m.toReturn) if sortSeries { diff --git a/storage/remote/azuread/azuread.go b/storage/remote/azuread/azuread.go index 20d48d0087..e2058fb54d 100644 --- a/storage/remote/azuread/azuread.go +++ b/storage/remote/azuread/azuread.go @@ -61,6 +61,12 @@ type OAuthConfig struct { TenantID string `yaml:"tenant_id,omitempty"` } +// SDKConfig is used to store azure SDK config values. +type SDKConfig struct { + // TenantID is the tenantId of the azure active directory application that is being used to authenticate. + TenantID string `yaml:"tenant_id,omitempty"` +} + // AzureADConfig is used to store the config values. type AzureADConfig struct { //nolint:revive // exported. // ManagedIdentity is the managed identity that is being used to authenticate. @@ -69,6 +75,9 @@ type AzureADConfig struct { //nolint:revive // exported. // OAuth is the oauth config that is being used to authenticate. OAuth *OAuthConfig `yaml:"oauth,omitempty"` + // OAuth is the oauth config that is being used to authenticate. + SDK *SDKConfig `yaml:"sdk,omitempty"` + // Cloud is the Azure cloud in which the service is running. Example: AzurePublic/AzureGovernment/AzureChina. Cloud string `yaml:"cloud,omitempty"` } @@ -102,14 +111,22 @@ func (c *AzureADConfig) Validate() error { return fmt.Errorf("must provide a cloud in the Azure AD config") } - if c.ManagedIdentity == nil && c.OAuth == nil { - return fmt.Errorf("must provide an Azure Managed Identity or Azure OAuth in the Azure AD config") + if c.ManagedIdentity == nil && c.OAuth == nil && c.SDK == nil { + return fmt.Errorf("must provide an Azure Managed Identity, Azure OAuth or Azure SDK in the Azure AD config") } if c.ManagedIdentity != nil && c.OAuth != nil { return fmt.Errorf("cannot provide both Azure Managed Identity and Azure OAuth in the Azure AD config") } + if c.ManagedIdentity != nil && c.SDK != nil { + return fmt.Errorf("cannot provide both Azure Managed Identity and Azure SDK in the Azure AD config") + } + + if c.OAuth != nil && c.SDK != nil { + return fmt.Errorf("cannot provide both Azure OAuth and Azure SDK in the Azure AD config") + } + if c.ManagedIdentity != nil { if c.ManagedIdentity.ClientID == "" { return fmt.Errorf("must provide an Azure Managed Identity client_id in the Azure AD config") @@ -143,6 +160,17 @@ func (c *AzureADConfig) Validate() error { } } + if c.SDK != nil { + var err error + + if c.SDK.TenantID != "" { + _, err = regexp.MatchString("^[0-9a-zA-Z-.]+$", c.SDK.TenantID) + if err != nil { + return fmt.Errorf("the provided Azure OAuth tenant_id is invalid") + } + } + } + return nil } @@ -225,6 +253,16 @@ func newTokenCredential(cfg *AzureADConfig) (azcore.TokenCredential, error) { } } + if cfg.SDK != nil { + sdkConfig := &SDKConfig{ + TenantID: cfg.SDK.TenantID, + } + cred, err = newSDKTokenCredential(clientOpts, sdkConfig) + if err != nil { + return nil, err + } + } + return cred, nil } @@ -241,6 +279,12 @@ func newOAuthTokenCredential(clientOpts *azcore.ClientOptions, oAuthConfig *OAut return azidentity.NewClientSecretCredential(oAuthConfig.TenantID, oAuthConfig.ClientID, oAuthConfig.ClientSecret, opts) } +// newSDKTokenCredential returns new SDK token credential. +func newSDKTokenCredential(clientOpts *azcore.ClientOptions, sdkConfig *SDKConfig) (azcore.TokenCredential, error) { + opts := &azidentity.DefaultAzureCredentialOptions{ClientOptions: *clientOpts, TenantID: sdkConfig.TenantID} + return azidentity.NewDefaultAzureCredential(opts) +} + // newTokenProvider helps to fetch accessToken for different types of credential. This also takes care of // refreshing the accessToken before expiry. This accessToken is attached to the Authorization header while making requests. func newTokenProvider(cfg *AzureADConfig, cred azcore.TokenCredential) (*tokenProvider, error) { diff --git a/storage/remote/azuread/azuread_test.go b/storage/remote/azuread/azuread_test.go index 5eed2c0b19..7c97138120 100644 --- a/storage/remote/azuread/azuread_test.go +++ b/storage/remote/azuread/azuread_test.go @@ -39,7 +39,7 @@ const ( testTokenString = "testTokenString" ) -var testTokenExpiry = time.Now().Add(5 * time.Second) +func testTokenExpiry() time.Time { return time.Now().Add(5 * time.Second) } type AzureAdTestSuite struct { suite.Suite @@ -94,7 +94,7 @@ func (ad *AzureAdTestSuite) TestAzureAdRoundTripper() { testToken := &azcore.AccessToken{ Token: testTokenString, - ExpiresOn: testTokenExpiry, + ExpiresOn: testTokenExpiry(), } ad.mockCredential.On("GetToken", mock.Anything, mock.Anything).Return(*testToken, nil) @@ -145,7 +145,7 @@ func TestAzureAdConfig(t *testing.T) { // Missing managedidentiy or oauth field. { filename: "testdata/azuread_bad_configmissing.yaml", - err: "must provide an Azure Managed Identity or Azure OAuth in the Azure AD config", + err: "must provide an Azure Managed Identity, Azure OAuth or Azure SDK in the Azure AD config", }, // Invalid managedidentity client id. { @@ -162,6 +162,11 @@ func TestAzureAdConfig(t *testing.T) { filename: "testdata/azuread_bad_twoconfig.yaml", err: "cannot provide both Azure Managed Identity and Azure OAuth in the Azure AD config", }, + // Invalid config when both sdk and oauth is provided. + { + filename: "testdata/azuread_bad_oauthsdkconfig.yaml", + err: "cannot provide both Azure OAuth and Azure SDK in the Azure AD config", + }, // Valid config with missing optionally cloud field. { filename: "testdata/azuread_good_cloudmissing.yaml", @@ -174,6 +179,10 @@ func TestAzureAdConfig(t *testing.T) { { filename: "testdata/azuread_good_oauth.yaml", }, + // Valid SDK config. + { + filename: "testdata/azuread_good_sdk.yaml", + }, } for _, c := range cases { _, err := loadAzureAdConfig(c.filename) @@ -232,6 +241,16 @@ func (s *TokenProviderTestSuite) TestNewTokenProvider() { }, err: "Cloud is not specified or is incorrect: ", }, + // Invalid tokenProvider for SDK. + { + cfg: &AzureADConfig{ + Cloud: "PublicAzure", + SDK: &SDKConfig{ + TenantID: dummyTenantID, + }, + }, + err: "Cloud is not specified or is incorrect: ", + }, // Valid tokenProvider for managedidentity. { cfg: &AzureADConfig{ @@ -252,6 +271,15 @@ func (s *TokenProviderTestSuite) TestNewTokenProvider() { }, }, }, + // Valid tokenProvider for SDK. + { + cfg: &AzureADConfig{ + Cloud: "AzurePublic", + SDK: &SDKConfig{ + TenantID: dummyTenantID, + }, + }, + }, } mockGetTokenCallCounter := 1 for _, c := range cases { @@ -264,11 +292,11 @@ func (s *TokenProviderTestSuite) TestNewTokenProvider() { } else { testToken := &azcore.AccessToken{ Token: testTokenString, - ExpiresOn: testTokenExpiry, + ExpiresOn: testTokenExpiry(), } s.mockCredential.On("GetToken", mock.Anything, mock.Anything).Return(*testToken, nil).Once(). - On("GetToken", mock.Anything, mock.Anything).Return(getToken(), nil) + On("GetToken", mock.Anything, mock.Anything).Return(getToken(), nil).Once() actualTokenProvider, actualErr := newTokenProvider(c.cfg, s.mockCredential) diff --git a/storage/remote/azuread/testdata/azuread_bad_oauthsdkconfig.yaml b/storage/remote/azuread/testdata/azuread_bad_oauthsdkconfig.yaml new file mode 100644 index 0000000000..825759d313 --- /dev/null +++ b/storage/remote/azuread/testdata/azuread_bad_oauthsdkconfig.yaml @@ -0,0 +1,7 @@ +cloud: AzurePublic +oauth: + client_id: 00000000-0000-0000-0000-000000000000 + client_secret: Cl1ent$ecret! + tenant_id: 00000000-a12b-3cd4-e56f-000000000000 +sdk: + tenant_id: 00000000-a12b-3cd4-e56f-000000000000 diff --git a/storage/remote/azuread/testdata/azuread_good_sdk.yaml b/storage/remote/azuread/testdata/azuread_good_sdk.yaml new file mode 100644 index 0000000000..53de8897d5 --- /dev/null +++ b/storage/remote/azuread/testdata/azuread_good_sdk.yaml @@ -0,0 +1,3 @@ +cloud: AzurePublic +sdk: + tenant_id: 00000000-a12b-3cd4-e56f-000000000000 diff --git a/storage/remote/client.go b/storage/remote/client.go index 5ba0f7117f..140194ec71 100644 --- a/storage/remote/client.go +++ b/storage/remote/client.go @@ -199,7 +199,7 @@ type RecoverableError struct { // Store sends a batch of samples to the HTTP endpoint, the request is the proto marshalled // and encoded bytes from codec.go. func (c *Client) Store(ctx context.Context, req []byte, attempt int) error { - httpReq, err := http.NewRequest("POST", c.urlString, bytes.NewReader(req)) + httpReq, err := http.NewRequest(http.MethodPost, c.urlString, bytes.NewReader(req)) if err != nil { // Errors from NewRequest are from unparsable URLs, so are not // recoverable. @@ -290,7 +290,7 @@ func (c *Client) Read(ctx context.Context, query *prompb.Query) (*prompb.QueryRe } compressed := snappy.Encode(nil, data) - httpReq, err := http.NewRequest("POST", c.urlString, bytes.NewReader(compressed)) + httpReq, err := http.NewRequest(http.MethodPost, c.urlString, bytes.NewReader(compressed)) if err != nil { return nil, fmt.Errorf("unable to create request: %w", err) } diff --git a/storage/remote/otlptranslator/README.md b/storage/remote/otlptranslator/README.md index c2b04e5aff..774fac5a7f 100644 --- a/storage/remote/otlptranslator/README.md +++ b/storage/remote/otlptranslator/README.md @@ -3,7 +3,6 @@ This files in the `prometheus/` and `prometheusremotewrite/` are copied from the OpenTelemetry Project[^1]. This is done instead of adding a go.mod dependency because OpenTelemetry depends on `prometheus/prometheus` and a cyclic dependency will be created. This is just a temporary solution and the long-term solution is to move the required packages from OpenTelemetry into `prometheus/prometheus`. -We don't copy in `./prometheus` through this script because that package imports a collector specific featuregate package we don't want to import. The featuregate package is being removed now, and in the future we will copy this folder too. To update the dependency is a multi-step process: 1. Vendor the latest `prometheus/prometheus`@`main` into [`opentelemetry/opentelemetry-collector-contrib`](https://github.com/open-telemetry/opentelemetry-collector-contrib) @@ -20,4 +19,4 @@ This means if we depend on the upstream packages directly, we will never able to When we do want to make changes to the types in `prompb`, we might need to edit the files directly. That is OK, please let @gouthamve or @jesusvazquez know so they can take care of updating the upstream code (by vendoring in `prometheus/prometheus` upstream and resolving conflicts) and then will run the copy script again to keep things updated. -[^1]: https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/pkg/translator/prometheus and https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/pkg/translator/prometheusremotewrite \ No newline at end of file +[^1]: https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/pkg/translator/prometheus and https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/pkg/translator/prometheusremotewrite diff --git a/storage/remote/otlptranslator/prometheus/normalize_label.go b/storage/remote/otlptranslator/prometheus/normalize_label.go index c02800c8bc..a6b41d1c37 100644 --- a/storage/remote/otlptranslator/prometheus/normalize_label.go +++ b/storage/remote/otlptranslator/prometheus/normalize_label.go @@ -3,7 +3,7 @@ // Copyright The OpenTelemetry Authors // SPDX-License-Identifier: Apache-2.0 -package prometheus // import "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/translator/prometheus" +package prometheus // import "github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheus" import ( "strings" diff --git a/storage/remote/otlptranslator/prometheus/normalize_name.go b/storage/remote/otlptranslator/prometheus/normalize_name.go index 7ae233a187..a976dfb485 100644 --- a/storage/remote/otlptranslator/prometheus/normalize_name.go +++ b/storage/remote/otlptranslator/prometheus/normalize_name.go @@ -3,7 +3,7 @@ // Copyright The OpenTelemetry Authors // SPDX-License-Identifier: Apache-2.0 -package prometheus // import "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/translator/prometheus" +package prometheus // import "github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheus" import ( "strings" diff --git a/storage/remote/otlptranslator/prometheus/unit_to_ucum.go b/storage/remote/otlptranslator/prometheus/unit_to_ucum.go index 4a72e683bb..718a520675 100644 --- a/storage/remote/otlptranslator/prometheus/unit_to_ucum.go +++ b/storage/remote/otlptranslator/prometheus/unit_to_ucum.go @@ -3,7 +3,7 @@ // Copyright The OpenTelemetry Authors // SPDX-License-Identifier: Apache-2.0 -package prometheus // import "github.com/open-telemetry/opentelemetry-collector-contrib/pkg/translator/prometheus" +package prometheus // import "github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheus" import "strings" diff --git a/storage/remote/otlptranslator/update-copy.sh b/storage/remote/otlptranslator/update-copy.sh index f90be5eedc..8aa645e0bd 100755 --- a/storage/remote/otlptranslator/update-copy.sh +++ b/storage/remote/otlptranslator/update-copy.sh @@ -23,5 +23,5 @@ case $(sed --help 2>&1) in *) set sed -i '';; esac -"$@" -e 's#github.com/open-telemetry/opentelemetry-collector-contrib/pkg/translator/prometheus#github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheus#g' ./prometheusremotewrite/*.go +"$@" -e 's#github.com/open-telemetry/opentelemetry-collector-contrib/pkg/translator/prometheus#github.com/prometheus/prometheus/storage/remote/otlptranslator/prometheus#g' ./prometheusremotewrite/*.go ./prometheus/*.go "$@" -e '1s#^#// DO NOT EDIT. COPIED AS-IS. SEE ../README.md\n\n#g' ./prometheusremotewrite/*.go ./prometheus/*.go diff --git a/storage/remote/read_handler.go b/storage/remote/read_handler.go index ffc64c9c3f..2a00ce897f 100644 --- a/storage/remote/read_handler.go +++ b/storage/remote/read_handler.go @@ -202,34 +202,16 @@ func (h *readHandler) remoteReadStreamedXORChunks(ctx context.Context, w http.Re return err } - querier, err := h.queryable.ChunkQuerier(query.StartTimestampMs, query.EndTimestampMs) - if err != nil { + chunks := h.getChunkSeriesSet(ctx, query, filteredMatchers) + if err := chunks.Err(); err != nil { return err } - defer func() { - if err := querier.Close(); err != nil { - level.Warn(h.logger).Log("msg", "Error on chunk querier close", "err", err.Error()) - } - }() - - var hints *storage.SelectHints - if query.Hints != nil { - hints = &storage.SelectHints{ - Start: query.Hints.StartMs, - End: query.Hints.EndMs, - Step: query.Hints.StepMs, - Func: query.Hints.Func, - Grouping: query.Hints.Grouping, - Range: query.Hints.RangeMs, - By: query.Hints.By, - } - } ws, err := StreamChunkedReadResponses( NewChunkedWriter(w, f), int64(i), // The streaming API has to provide the series sorted. - querier.Select(ctx, true, hints, filteredMatchers...), + chunks, sortedExternalLabels, h.remoteReadMaxBytesInFrame, h.marshalPool, @@ -254,6 +236,35 @@ func (h *readHandler) remoteReadStreamedXORChunks(ctx context.Context, w http.Re } } +// getChunkSeriesSet executes a query to retrieve a ChunkSeriesSet, +// encapsulating the operation in its own function to ensure timely release of +// the querier resources. +func (h *readHandler) getChunkSeriesSet(ctx context.Context, query *prompb.Query, filteredMatchers []*labels.Matcher) storage.ChunkSeriesSet { + querier, err := h.queryable.ChunkQuerier(query.StartTimestampMs, query.EndTimestampMs) + if err != nil { + return storage.ErrChunkSeriesSet(err) + } + defer func() { + if err := querier.Close(); err != nil { + level.Warn(h.logger).Log("msg", "Error on chunk querier close", "err", err.Error()) + } + }() + + var hints *storage.SelectHints + if query.Hints != nil { + hints = &storage.SelectHints{ + Start: query.Hints.StartMs, + End: query.Hints.EndMs, + Step: query.Hints.StepMs, + Func: query.Hints.Func, + Grouping: query.Hints.Grouping, + Range: query.Hints.RangeMs, + By: query.Hints.By, + } + } + return querier.Select(ctx, true, hints, filteredMatchers...) +} + // filterExtLabelsFromMatchers change equality matchers which match external labels // to a matcher that looks for an empty label, // as that label should not be present in the storage. diff --git a/storage/remote/read_handler_test.go b/storage/remote/read_handler_test.go index e83a0cb21a..e8e0ecb8df 100644 --- a/storage/remote/read_handler_test.go +++ b/storage/remote/read_handler_test.go @@ -75,7 +75,7 @@ func TestSampledReadEndpoint(t *testing.T) { require.NoError(t, err) compressed := snappy.Encode(nil, data) - request, err := http.NewRequest("POST", "", bytes.NewBuffer(compressed)) + request, err := http.NewRequest(http.MethodPost, "", bytes.NewBuffer(compressed)) require.NoError(t, err) recorder := httptest.NewRecorder() @@ -170,7 +170,7 @@ func BenchmarkStreamReadEndpoint(b *testing.B) { for i := 0; i < b.N; i++ { compressed := snappy.Encode(nil, data) - request, err := http.NewRequest("POST", "", bytes.NewBuffer(compressed)) + request, err := http.NewRequest(http.MethodPost, "", bytes.NewBuffer(compressed)) require.NoError(b, err) recorder := httptest.NewRecorder() @@ -268,7 +268,7 @@ func TestStreamReadEndpoint(t *testing.T) { require.NoError(t, err) compressed := snappy.Encode(nil, data) - request, err := http.NewRequest("POST", "", bytes.NewBuffer(compressed)) + request, err := http.NewRequest(http.MethodPost, "", bytes.NewBuffer(compressed)) require.NoError(t, err) recorder := httptest.NewRecorder() diff --git a/storage/remote/write_test.go b/storage/remote/write_test.go index 4a9a3bafc4..c79ac3ab7d 100644 --- a/storage/remote/write_test.go +++ b/storage/remote/write_test.go @@ -409,7 +409,7 @@ func generateOTLPWriteRequest(t *testing.T) pmetricotlp.ExportRequest { // Generate One Counter, One Gauge, One Histogram, One Exponential-Histogram // with resource attributes: service.name="test-service", service.instance.id="test-instance", host.name="test-host" - // with metric attibute: foo.bar="baz" + // with metric attribute: foo.bar="baz" timestamp := time.Now() diff --git a/tsdb/block_test.go b/tsdb/block_test.go index ad7eb55575..21e20a61c8 100644 --- a/tsdb/block_test.go +++ b/tsdb/block_test.go @@ -209,6 +209,22 @@ func TestCorruptedChunk(t *testing.T) { } } +func sequenceFiles(dir string) ([]string, error) { + files, err := os.ReadDir(dir) + if err != nil { + return nil, err + } + var res []string + + for _, fi := range files { + if _, err := strconv.ParseUint(fi.Name(), 10, 64); err != nil { + continue + } + res = append(res, filepath.Join(dir, fi.Name())) + } + return res, nil +} + func TestLabelValuesWithMatchers(t *testing.T) { tmpdir := t.TempDir() ctx := context.Background() diff --git a/tsdb/chunks/chunks.go b/tsdb/chunks/chunks.go index 543b98c289..0826f69670 100644 --- a/tsdb/chunks/chunks.go +++ b/tsdb/chunks/chunks.go @@ -202,15 +202,6 @@ func ChunkFromSamplesGeneric(s Samples) (Meta, error) { }, nil } -// PopulatedChunk creates a chunk populated with samples every second starting at minTime. -func PopulatedChunk(numSamples int, minTime int64) (Meta, error) { - samples := make([]Sample, numSamples) - for i := 0; i < numSamples; i++ { - samples[i] = sample{t: minTime + int64(i*1000), f: 1.0} - } - return ChunkFromSamples(samples) -} - // ChunkMetasToSamples converts a slice of chunk meta data to a slice of samples. // Used in tests to compare the content of chunks. func ChunkMetasToSamples(chunks []Meta) (result []Sample) { diff --git a/tsdb/chunks/head_chunks.go b/tsdb/chunks/head_chunks.go index 5ba5381320..087f25fbb3 100644 --- a/tsdb/chunks/head_chunks.go +++ b/tsdb/chunks/head_chunks.go @@ -15,7 +15,6 @@ package chunks import ( "bufio" - "bytes" "encoding/binary" "errors" "fmt" @@ -690,7 +689,6 @@ func (cdm *ChunkDiskMapper) Chunk(ref ChunkDiskMapperRef) (chunkenc.Chunk, error sgmIndex, chkStart := ref.Unpack() // We skip the series ref and the mint/maxt beforehand. chkStart += SeriesRefSize + (2 * MintMaxtSize) - chkCRC32 := newCRC32() // If it is the current open file, then the chunks can be in the buffer too. if sgmIndex == cdm.curFileSequence { @@ -755,20 +753,13 @@ func (cdm *ChunkDiskMapper) Chunk(ref ChunkDiskMapperRef) (chunkenc.Chunk, error // Check the CRC. sum := mmapFile.byteSlice.Range(chkDataEnd, chkDataEnd+CRCSize) - if _, err := chkCRC32.Write(mmapFile.byteSlice.Range(chkStart-(SeriesRefSize+2*MintMaxtSize), chkDataEnd)); err != nil { + if err := checkCRC32(mmapFile.byteSlice.Range(chkStart-(SeriesRefSize+2*MintMaxtSize), chkDataEnd), sum); err != nil { return nil, &CorruptionErr{ Dir: cdm.dir.Name(), FileIndex: sgmIndex, Err: err, } } - if act := chkCRC32.Sum(nil); !bytes.Equal(act, sum) { - return nil, &CorruptionErr{ - Dir: cdm.dir.Name(), - FileIndex: sgmIndex, - Err: fmt.Errorf("checksum mismatch expected:%x, actual:%x", sum, act), - } - } // The chunk data itself. chkData := mmapFile.byteSlice.Range(chkDataEnd-int(chkDataLen), chkDataEnd) @@ -802,8 +793,6 @@ func (cdm *ChunkDiskMapper) IterateAllChunks(f func(seriesRef HeadSeriesRef, chu cdm.fileMaxtSet = true }() - chkCRC32 := newCRC32() - // Iterate files in ascending order. segIDs := make([]int, 0, len(cdm.mmappedChunkFiles)) for seg := range cdm.mmappedChunkFiles { @@ -838,7 +827,6 @@ func (cdm *ChunkDiskMapper) IterateAllChunks(f func(seriesRef HeadSeriesRef, chu " - required:%v, available:%v, file:%d", idx+MaxHeadChunkMetaSize, fileEnd, segID), } } - chkCRC32.Reset() chunkRef := newChunkDiskMapperRef(uint64(segID), uint64(idx)) startIdx := idx @@ -877,14 +865,11 @@ func (cdm *ChunkDiskMapper) IterateAllChunks(f func(seriesRef HeadSeriesRef, chu // Check CRC. sum := mmapFile.byteSlice.Range(idx, idx+CRCSize) - if _, err := chkCRC32.Write(mmapFile.byteSlice.Range(startIdx, idx)); err != nil { - return err - } - if act := chkCRC32.Sum(nil); !bytes.Equal(act, sum) { + if err := checkCRC32(mmapFile.byteSlice.Range(startIdx, idx), sum); err != nil { return &CorruptionErr{ Dir: cdm.dir.Name(), FileIndex: segID, - Err: fmt.Errorf("checksum mismatch expected:%x, actual:%x", sum, act), + Err: err, } } idx += CRCSize diff --git a/tsdb/compact.go b/tsdb/compact.go index 3d8d9130c4..e09039cf33 100644 --- a/tsdb/compact.go +++ b/tsdb/compact.go @@ -60,7 +60,7 @@ type Compactor interface { // Write persists a Block into a directory. // No Block is written when resulting Block has 0 samples, and returns empty ulid.ULID{}. - Write(dest string, b BlockReader, mint, maxt int64, parent *BlockMeta) (ulid.ULID, error) + Write(dest string, b BlockReader, mint, maxt int64, base *BlockMeta) (ulid.ULID, error) // Compact runs compaction against the provided directories. Must // only be called concurrently with results of Plan(). @@ -96,7 +96,8 @@ type CompactorMetrics struct { ChunkRange prometheus.Histogram } -func newCompactorMetrics(r prometheus.Registerer) *CompactorMetrics { +// NewCompactorMetrics initializes metrics for Compactor. +func NewCompactorMetrics(r prometheus.Registerer) *CompactorMetrics { m := &CompactorMetrics{} m.Ran = prometheus.NewCounter(prometheus.CounterOpts{ @@ -203,7 +204,7 @@ func NewLeveledCompactorWithOptions(ctx context.Context, r prometheus.Registerer ranges: ranges, chunkPool: pool, logger: l, - metrics: newCompactorMetrics(r), + metrics: NewCompactorMetrics(r), ctx: ctx, maxBlockChunkSegmentSize: maxBlockChunkSegmentSize, mergeFunc: mergeFunc, @@ -535,7 +536,7 @@ func (c *LeveledCompactor) CompactWithBlockPopulator(dest string, dirs []string, return uid, errs.Err() } -func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64, parent *BlockMeta) (ulid.ULID, error) { +func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64, base *BlockMeta) (ulid.ULID, error) { start := time.Now() uid := ulid.MustNew(ulid.Now(), rand.Reader) @@ -548,9 +549,12 @@ func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64, p meta.Compaction.Level = 1 meta.Compaction.Sources = []ulid.ULID{uid} - if parent != nil { + if base != nil { meta.Compaction.Parents = []BlockDesc{ - {ULID: parent.ULID, MinTime: parent.MinTime, MaxTime: parent.MaxTime}, + {ULID: base.ULID, MinTime: base.MinTime, MaxTime: base.MaxTime}, + } + if base.Compaction.FromOutOfOrder() { + meta.Compaction.SetOutOfOrder() } } @@ -575,6 +579,7 @@ func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64, p "maxt", meta.MaxTime, "ulid", meta.ULID, "duration", time.Since(start), + "ooo", meta.Compaction.FromOutOfOrder(), ) return uid, nil } diff --git a/tsdb/db.go b/tsdb/db.go index 4998da6aa7..d675635d7a 100644 --- a/tsdb/db.go +++ b/tsdb/db.go @@ -24,7 +24,6 @@ import ( "os" "path/filepath" "slices" - "strconv" "strings" "sync" "time" @@ -43,7 +42,7 @@ import ( "github.com/prometheus/prometheus/tsdb/chunks" tsdb_errors "github.com/prometheus/prometheus/tsdb/errors" "github.com/prometheus/prometheus/tsdb/fileutil" - _ "github.com/prometheus/prometheus/tsdb/goversion" // Load the package into main to make sure minium Go version is met. + _ "github.com/prometheus/prometheus/tsdb/goversion" // Load the package into main to make sure minimum Go version is met. "github.com/prometheus/prometheus/tsdb/tsdbutil" "github.com/prometheus/prometheus/tsdb/wlog" ) @@ -530,9 +529,10 @@ func (db *DBReadOnly) loadDataAsQueryable(maxt int64) (storage.SampleAndChunkQue if err := head.Init(maxBlockTime); err != nil { return nil, fmt.Errorf("read WAL: %w", err) } - // Set the wal to nil to disable all wal operations. + // Set the wal and the wbl to nil to disable related operations. // This is mainly to avoid blocking when closing the head. head.wal = nil + head.wbl = nil } db.closers = append(db.closers, head) @@ -779,10 +779,6 @@ func open(dir string, l log.Logger, r prometheus.Registerer, opts *Options, rngs walDir := filepath.Join(dir, "wal") wblDir := filepath.Join(dir, wlog.WblDirName) - // Migrate old WAL if one exists. - if err := MigrateWAL(l, walDir); err != nil { - return nil, fmt.Errorf("migrate WAL: %w", err) - } for _, tmpDir := range []string{walDir, dir} { // Remove tmp dirs. if err := removeBestEffortTmpDirs(l, tmpDir); err != nil { @@ -1303,25 +1299,17 @@ func (db *DB) compactOOO(dest string, oooHead *OOOCompactionHead) (_ []ulid.ULID } }() + meta := &BlockMeta{} + meta.Compaction.SetOutOfOrder() for t := blockSize * (oooHeadMint / blockSize); t <= oooHeadMaxt; t += blockSize { mint, maxt := t, t+blockSize // Block intervals are half-open: [b.MinTime, b.MaxTime). Block intervals are always +1 than the total samples it includes. - uid, err := db.compactor.Write(dest, oooHead.CloneForTimeRange(mint, maxt-1), mint, maxt, nil) + uid, err := db.compactor.Write(dest, oooHead.CloneForTimeRange(mint, maxt-1), mint, maxt, meta) if err != nil { return nil, err } if uid.Compare(ulid.ULID{}) != 0 { ulids = append(ulids, uid) - blockDir := filepath.Join(dest, uid.String()) - meta, _, err := readMetaFile(blockDir) - if err != nil { - return ulids, fmt.Errorf("read meta: %w", err) - } - meta.Compaction.SetOutOfOrder() - _, err = writeMetaFile(db.logger, blockDir, meta) - if err != nil { - return ulids, fmt.Errorf("write meta: %w", err) - } } } @@ -1369,6 +1357,14 @@ func (db *DB) compactHead(head *RangeHead) error { func (db *DB) compactBlocks() (err error) { // Check for compactions of multiple blocks. for { + // If we have a lot of blocks to compact the whole process might take + // long enough that we end up with a HEAD block that needs to be written. + // Check if that's the case and stop compactions early. + if db.head.compactable() { + level.Warn(db.logger).Log("msg", "aborting block compactions to persit the head block") + return nil + } + plan, err := db.compactor.Plan(db.dir) if err != nil { return fmt.Errorf("plan compaction: %w", err) @@ -1613,9 +1609,9 @@ func BeyondTimeRetention(db *DB, blocks []*Block) (deletable map[ulid.ULID]struc deletable = make(map[ulid.ULID]struct{}) for i, block := range blocks { - // The difference between the first block and this block is larger than + // The difference between the first block and this block is greater than or equal to // the retention period so any blocks after that are added as deletable. - if i > 0 && blocks[0].Meta().MaxTime-block.Meta().MaxTime > db.opts.RetentionDuration { + if i > 0 && blocks[0].Meta().MaxTime-block.Meta().MaxTime >= db.opts.RetentionDuration { for _, b := range blocks[i:] { deletable[b.meta.ULID] = struct{}{} } @@ -2213,39 +2209,6 @@ func blockDirs(dir string) ([]string, error) { return dirs, nil } -func sequenceFiles(dir string) ([]string, error) { - files, err := os.ReadDir(dir) - if err != nil { - return nil, err - } - var res []string - - for _, fi := range files { - if _, err := strconv.ParseUint(fi.Name(), 10, 64); err != nil { - continue - } - res = append(res, filepath.Join(dir, fi.Name())) - } - return res, nil -} - -func nextSequenceFile(dir string) (string, int, error) { - files, err := os.ReadDir(dir) - if err != nil { - return "", 0, err - } - - i := uint64(0) - for _, f := range files { - j, err := strconv.ParseUint(f.Name(), 10, 64) - if err != nil { - continue - } - i = j - } - return filepath.Join(dir, fmt.Sprintf("%0.6d", i+1)), int(i + 1), nil -} - func exponential(d, min, max time.Duration) time.Duration { d *= 2 if d < min { diff --git a/tsdb/db_test.go b/tsdb/db_test.go index 45a16b8ba3..71b2f05ac7 100644 --- a/tsdb/db_test.go +++ b/tsdb/db_test.go @@ -1441,7 +1441,7 @@ func (c *mockCompactorFailing) Write(dest string, _ BlockReader, _, _ int64, _ * c.blocks = append(c.blocks, block) // Now check that all expected blocks are actually persisted on disk. - // This way we make sure that the we have some blocks that are supposed to be removed. + // This way we make sure that we have some blocks that are supposed to be removed. var expectedBlocks []string for _, b := range c.blocks { expectedBlocks = append(expectedBlocks, filepath.Join(dest, b.Meta().ULID.String())) @@ -1463,34 +1463,66 @@ func (*mockCompactorFailing) CompactOOO(string, *OOOCompactionHead) (result []ul } func TestTimeRetention(t *testing.T) { - db := openTestDB(t, nil, []int64{1000}) - defer func() { - require.NoError(t, db.Close()) - }() - - blocks := []*BlockMeta{ - {MinTime: 500, MaxTime: 900}, // Oldest block - {MinTime: 1000, MaxTime: 1500}, - {MinTime: 1500, MaxTime: 2000}, // Newest Block + testCases := []struct { + name string + blocks []*BlockMeta + expBlocks []*BlockMeta + retentionDuration int64 + }{ + { + name: "Block max time delta greater than retention duration", + blocks: []*BlockMeta{ + {MinTime: 500, MaxTime: 900}, // Oldest block, beyond retention + {MinTime: 1000, MaxTime: 1500}, + {MinTime: 1500, MaxTime: 2000}, // Newest block + }, + expBlocks: []*BlockMeta{ + {MinTime: 1000, MaxTime: 1500}, + {MinTime: 1500, MaxTime: 2000}, + }, + retentionDuration: 1000, + }, + { + name: "Block max time delta equal to retention duration", + blocks: []*BlockMeta{ + {MinTime: 500, MaxTime: 900}, // Oldest block + {MinTime: 1000, MaxTime: 1500}, // Coinciding exactly with the retention duration. + {MinTime: 1500, MaxTime: 2000}, // Newest block + }, + expBlocks: []*BlockMeta{ + {MinTime: 1500, MaxTime: 2000}, + }, + retentionDuration: 500, + }, } + for _, tc := range testCases { + t.Run(tc.name, func(t *testing.T) { + db := openTestDB(t, nil, []int64{1000}) + defer func() { + require.NoError(t, db.Close()) + }() - for _, m := range blocks { - createBlock(t, db.Dir(), genSeries(10, 10, m.MinTime, m.MaxTime)) + for _, m := range tc.blocks { + createBlock(t, db.Dir(), genSeries(10, 10, m.MinTime, m.MaxTime)) + } + + require.NoError(t, db.reloadBlocks()) // Reload the db to register the new blocks. + require.Len(t, db.Blocks(), len(tc.blocks)) // Ensure all blocks are registered. + + db.opts.RetentionDuration = tc.retentionDuration + // Reloading should truncate the blocks which are >= the retention duration vs the first block. + require.NoError(t, db.reloadBlocks()) + + actBlocks := db.Blocks() + + require.Equal(t, 1, int(prom_testutil.ToFloat64(db.metrics.timeRetentionCount)), "metric retention count mismatch") + require.Len(t, actBlocks, len(tc.expBlocks)) + for i, eb := range tc.expBlocks { + require.Equal(t, eb.MinTime, actBlocks[i].meta.MinTime) + require.Equal(t, eb.MaxTime, actBlocks[i].meta.MaxTime) + } + }) } - - require.NoError(t, db.reloadBlocks()) // Reload the db to register the new blocks. - require.Equal(t, len(blocks), len(db.Blocks())) // Ensure all blocks are registered. - - db.opts.RetentionDuration = blocks[2].MaxTime - blocks[1].MinTime - require.NoError(t, db.reloadBlocks()) - - expBlocks := blocks[1:] - actBlocks := db.Blocks() - - require.Equal(t, 1, int(prom_testutil.ToFloat64(db.metrics.timeRetentionCount)), "metric retention count mismatch") - require.Equal(t, len(expBlocks), len(actBlocks)) - require.Equal(t, expBlocks[0].MaxTime, actBlocks[0].meta.MaxTime) - require.Equal(t, expBlocks[len(expBlocks)-1].MaxTime, actBlocks[len(actBlocks)-1].meta.MaxTime) } func TestRetentionDurationMetric(t *testing.T) { @@ -1966,6 +1998,7 @@ func TestInitializeHeadTimestamp(t *testing.T) { // Should be set to init values if no WAL or blocks exist so far. require.Equal(t, int64(math.MaxInt64), db.head.MinTime()) require.Equal(t, int64(math.MinInt64), db.head.MaxTime()) + require.False(t, db.head.initialized()) // First added sample initializes the writable range. ctx := context.Background() @@ -1975,6 +2008,7 @@ func TestInitializeHeadTimestamp(t *testing.T) { require.Equal(t, int64(1000), db.head.MinTime()) require.Equal(t, int64(1000), db.head.MaxTime()) + require.True(t, db.head.initialized()) }) t.Run("wal-only", func(t *testing.T) { dir := t.TempDir() @@ -2003,6 +2037,7 @@ func TestInitializeHeadTimestamp(t *testing.T) { require.Equal(t, int64(5000), db.head.MinTime()) require.Equal(t, int64(15000), db.head.MaxTime()) + require.True(t, db.head.initialized()) }) t.Run("existing-block", func(t *testing.T) { dir := t.TempDir() @@ -2015,6 +2050,7 @@ func TestInitializeHeadTimestamp(t *testing.T) { require.Equal(t, int64(2000), db.head.MinTime()) require.Equal(t, int64(2000), db.head.MaxTime()) + require.True(t, db.head.initialized()) }) t.Run("existing-block-and-wal", func(t *testing.T) { dir := t.TempDir() @@ -2047,6 +2083,7 @@ func TestInitializeHeadTimestamp(t *testing.T) { require.Equal(t, int64(6000), db.head.MinTime()) require.Equal(t, int64(15000), db.head.MaxTime()) + require.True(t, db.head.initialized()) // Check that old series has been GCed. require.Equal(t, 1.0, prom_testutil.ToFloat64(db.head.metrics.series)) }) @@ -3598,7 +3635,7 @@ func testChunkQuerierShouldNotPanicIfHeadChunkIsTruncatedWhileReadingQueriedChun // just to iterate through the bytes slice. We don't really care the reason why // we read this data, we just need to read it to make sure the memory address // of the []byte is still valid. - chkCRC32 := newCRC32() + chkCRC32 := crc32.New(crc32.MakeTable(crc32.Castagnoli)) for _, chunk := range chunks { chkCRC32.Reset() _, err := chkCRC32.Write(chunk.Bytes()) @@ -6949,3 +6986,63 @@ func requireEqualOOOSamples(t *testing.T, expectedSamples int, db *DB) { prom_testutil.ToFloat64(db.head.metrics.outOfOrderSamplesAppended.WithLabelValues(sampleMetricTypeFloat)), "number of ooo appended samples mismatch") } + +type mockCompactorFn struct { + planFn func() ([]string, error) + compactFn func() (ulid.ULID, error) + writeFn func() (ulid.ULID, error) +} + +func (c *mockCompactorFn) Plan(_ string) ([]string, error) { + return c.planFn() +} + +func (c *mockCompactorFn) Compact(_ string, _ []string, _ []*Block) (ulid.ULID, error) { + return c.compactFn() +} + +func (c *mockCompactorFn) Write(_ string, _ BlockReader, _, _ int64, _ *BlockMeta) (ulid.ULID, error) { + return c.writeFn() +} + +// Regression test for https://github.com/prometheus/prometheus/pull/13754 +func TestAbortBlockCompactions(t *testing.T) { + // Create a test DB + db := openTestDB(t, nil, nil) + defer func() { + require.NoError(t, db.Close()) + }() + // It should NOT be compactible at the beginning of the test + require.False(t, db.head.compactable(), "head should NOT be compactable") + + // Track the number of compactions run inside db.compactBlocks() + var compactions int + + // Use a mock compactor with custom Plan() implementation + db.compactor = &mockCompactorFn{ + planFn: func() ([]string, error) { + // On every Plan() run increment compactions. After 4 compactions + // update HEAD to make it compactible to force an exit from db.compactBlocks() loop. + compactions++ + if compactions > 3 { + chunkRange := db.head.chunkRange.Load() + db.head.minTime.Store(0) + db.head.maxTime.Store(chunkRange * 2) + require.True(t, db.head.compactable(), "head should be compactable") + } + // Our custom Plan() will always return something to compact. + return []string{"1", "2", "3"}, nil + }, + compactFn: func() (ulid.ULID, error) { + return ulid.ULID{}, nil + }, + writeFn: func() (ulid.ULID, error) { + return ulid.ULID{}, nil + }, + } + + err := db.Compact(context.Background()) + require.NoError(t, err) + require.True(t, db.head.compactable(), "head should be compactable") + require.Equal(t, 4, compactions, "expected 4 compactions to be completed") +} diff --git a/tsdb/head.go b/tsdb/head.go index 7e6fda9c7e..8b3d9787ca 100644 --- a/tsdb/head.go +++ b/tsdb/head.go @@ -620,6 +620,7 @@ func (h *Head) Init(minValidTime int64) error { refSeries := make(map[chunks.HeadSeriesRef]*memSeries) snapshotLoaded := false + var chunkSnapshotLoadDuration time.Duration if h.opts.EnableMemorySnapshotOnShutdown { level.Info(h.logger).Log("msg", "Chunk snapshot is enabled, replaying from the snapshot") // If there are any WAL files, there should be at least one WAL file with an index that is current or newer @@ -650,7 +651,8 @@ func (h *Head) Init(minValidTime int64) error { snapIdx, snapOffset, refSeries, err = h.loadChunkSnapshot() if err == nil { snapshotLoaded = true - level.Info(h.logger).Log("msg", "Chunk snapshot loading time", "duration", time.Since(start).String()) + chunkSnapshotLoadDuration = time.Since(start) + level.Info(h.logger).Log("msg", "Chunk snapshot loading time", "duration", chunkSnapshotLoadDuration.String()) } if err != nil { snapIdx, snapOffset = -1, 0 @@ -672,6 +674,8 @@ func (h *Head) Init(minValidTime int64) error { oooMmappedChunks map[chunks.HeadSeriesRef][]*mmappedChunk lastMmapRef chunks.ChunkDiskMapperRef err error + + mmapChunkReplayDuration time.Duration ) if snapshotLoaded || h.wal != nil { // If snapshot was not loaded and if there is no WAL, then m-map chunks will be discarded @@ -695,7 +699,8 @@ func (h *Head) Init(minValidTime int64) error { return err } } - level.Info(h.logger).Log("msg", "On-disk memory mappable chunks replay completed", "duration", time.Since(mmapChunkReplayStart).String()) + mmapChunkReplayDuration = time.Since(mmapChunkReplayStart) + level.Info(h.logger).Log("msg", "On-disk memory mappable chunks replay completed", "duration", mmapChunkReplayDuration.String()) } if h.wal == nil { @@ -817,6 +822,8 @@ func (h *Head) Init(minValidTime int64) error { "checkpoint_replay_duration", checkpointReplayDuration.String(), "wal_replay_duration", walReplayDuration.String(), "wbl_replay_duration", wblReplayDuration.String(), + "chunk_snapshot_load_duration", chunkSnapshotLoadDuration.String(), + "mmap_chunk_replay_duration", mmapChunkReplayDuration.String(), "total_replay_duration", totalReplayDuration.String(), ) @@ -1074,11 +1081,11 @@ func (h *Head) SetMinValidTime(minValidTime int64) { // Truncate removes old data before mint from the head and WAL. func (h *Head) Truncate(mint int64) (err error) { - initialize := h.MinTime() == math.MaxInt64 + initialized := h.initialized() if err := h.truncateMemory(mint); err != nil { return err } - if initialize { + if !initialized { return nil } return h.truncateWAL(mint) @@ -1100,9 +1107,9 @@ func (h *Head) truncateMemory(mint int64) (err error) { } }() - initialize := h.MinTime() == math.MaxInt64 + initialized := h.initialized() - if h.MinTime() >= mint && !initialize { + if h.MinTime() >= mint && initialized { return nil } @@ -1113,7 +1120,7 @@ func (h *Head) truncateMemory(mint int64) (err error) { defer h.memTruncationInProcess.Store(false) // We wait for pending queries to end that overlap with this truncation. - if !initialize { + if initialized { h.WaitForPendingReadersInTimeRange(h.MinTime(), mint) } @@ -1127,7 +1134,7 @@ func (h *Head) truncateMemory(mint int64) (err error) { // This was an initial call to Truncate after loading blocks on startup. // We haven't read back the WAL yet, so do not attempt to truncate it. - if initialize { + if !initialized { return nil } @@ -1615,10 +1622,19 @@ func (h *Head) MaxOOOTime() int64 { return h.maxOOOTime.Load() } +// initialized returns true if the head has a MinTime set, false otherwise. +func (h *Head) initialized() bool { + return h.MinTime() != math.MaxInt64 +} + // compactable returns whether the head has a compactable range. // The head has a compactable range when the head time range is 1.5 times the chunk range. // The 0.5 acts as a buffer of the appendable window. func (h *Head) compactable() bool { + if !h.initialized() { + return false + } + return h.MaxTime()-h.MinTime() > h.chunkRange.Load()/2*3 } diff --git a/tsdb/head_append.go b/tsdb/head_append.go index 6342a19d46..23c2c0fbd7 100644 --- a/tsdb/head_append.go +++ b/tsdb/head_append.go @@ -138,7 +138,7 @@ func (h *Head) Appender(_ context.Context) storage.Appender { // The head cache might not have a starting point yet. The init appender // picks up the first appended timestamp as the base. - if h.MinTime() == math.MaxInt64 { + if !h.initialized() { return &initAppender{ head: h, } @@ -191,20 +191,13 @@ func (h *Head) appendableMinValidTime() int64 { // AppendableMinValidTime returns the minimum valid time for samples to be appended to the Head. // Returns false if Head hasn't been initialized yet and the minimum time isn't known yet. func (h *Head) AppendableMinValidTime() (int64, bool) { - if h.MinTime() == math.MaxInt64 { + if !h.initialized() { return 0, false } return h.appendableMinValidTime(), true } -func max(a, b int64) int64 { - if a > b { - return a - } - return b -} - func (h *Head) getAppendBuffer() []record.RefSample { b := h.appendPool.Get() if b == nil { diff --git a/tsdb/head_read.go b/tsdb/head_read.go index 10a4623924..45bbc81f18 100644 --- a/tsdb/head_read.go +++ b/tsdb/head_read.go @@ -582,17 +582,6 @@ func (s *memSeries) oooMergedChunks(meta chunks.Meta, cdm *chunks.ChunkDiskMappe return mc, nil } -var _ chunkenc.Iterable = &mergedOOOChunks{} - -// mergedOOOChunks holds the list of iterables for overlapping chunks. -type mergedOOOChunks struct { - chunkIterables []chunkenc.Iterable -} - -func (o mergedOOOChunks) Iterator(iterator chunkenc.Iterator) chunkenc.Iterator { - return storage.ChainSampleIteratorFromIterables(iterator, o.chunkIterables) -} - var _ chunkenc.Iterable = &boundedIterable{} // boundedIterable is an implementation of chunkenc.Iterable that uses a diff --git a/tsdb/head_test.go b/tsdb/head_test.go index 41c2e062f2..750c8a11e5 100644 --- a/tsdb/head_test.go +++ b/tsdb/head_test.go @@ -5819,3 +5819,16 @@ func TestHeadAppender_AppendCTZeroSample(t *testing.T) { require.Equal(t, chunkenc.ValNone, it.Next()) } } + +func TestHeadCompactableDoesNotCompactEmptyHead(t *testing.T) { + // Use a chunk range of 1 here so that if we attempted to determine if the head + // was compactable using default values for min and max times, `Head.compactable()` + // would return true which is incorrect. This test verifies that we short-circuit + // the check when the head has not yet had any samples added. + head, _ := newTestHead(t, 1, wlog.CompressionNone, false) + defer func() { + require.NoError(t, head.Close()) + }() + + require.False(t, head.compactable()) +} diff --git a/tsdb/index/index.go b/tsdb/index/index.go index 7aae4c2645..7ab890b99e 100644 --- a/tsdb/index/index.go +++ b/tsdb/index/index.go @@ -1829,7 +1829,7 @@ func NewStringListIter(s []string) StringIter { return &stringListIter{l: s} } -// symbolsIter implements StringIter. +// stringListIter implements StringIter. type stringListIter struct { l []string cur string diff --git a/tsdb/ooo_head_read.go b/tsdb/ooo_head_read.go index c9fe5cd580..ed0b3fd227 100644 --- a/tsdb/ooo_head_read.go +++ b/tsdb/ooo_head_read.go @@ -42,6 +42,17 @@ type OOOHeadIndexReader struct { lastGarbageCollectedMmapRef chunks.ChunkDiskMapperRef } +var _ chunkenc.Iterable = &mergedOOOChunks{} + +// mergedOOOChunks holds the list of iterables for overlapping chunks. +type mergedOOOChunks struct { + chunkIterables []chunkenc.Iterable +} + +func (o mergedOOOChunks) Iterator(iterator chunkenc.Iterator) chunkenc.Iterator { + return storage.ChainSampleIteratorFromIterables(iterator, o.chunkIterables) +} + func NewOOOHeadIndexReader(head *Head, mint, maxt int64, lastGarbageCollectedMmapRef chunks.ChunkDiskMapperRef) *OOOHeadIndexReader { hr := &headIndexReader{ head: head, diff --git a/tsdb/querier.go b/tsdb/querier.go index 6fc2758dbe..2ee7f5153b 100644 --- a/tsdb/querier.go +++ b/tsdb/querier.go @@ -19,8 +19,6 @@ import ( "fmt" "math" "slices" - "strings" - "unicode/utf8" "github.com/oklog/ulid" @@ -35,20 +33,6 @@ import ( "github.com/prometheus/prometheus/util/annotations" ) -// Bitmap used by func isRegexMetaCharacter to check whether a character needs to be escaped. -var regexMetaCharacterBytes [16]byte - -// isRegexMetaCharacter reports whether byte b needs to be escaped. -func isRegexMetaCharacter(b byte) bool { - return b < utf8.RuneSelf && regexMetaCharacterBytes[b%16]&(1<<(b/16)) != 0 -} - -func init() { - for _, b := range []byte(`.+*?()|[]{}^$`) { - regexMetaCharacterBytes[b%16] |= 1 << (b / 16) - } -} - type blockBaseQuerier struct { blockID ulid.ULID index IndexReader @@ -195,55 +179,6 @@ func (q *blockChunkQuerier) Select(ctx context.Context, sortSeries bool, hints * return NewBlockChunkSeriesSet(q.blockID, q.index, q.chunks, q.tombstones, p, mint, maxt, disableTrimming) } -func findSetMatches(pattern string) []string { - // Return empty matches if the wrapper from Prometheus is missing. - if len(pattern) < 6 || pattern[:4] != "^(?:" || pattern[len(pattern)-2:] != ")$" { - return nil - } - escaped := false - sets := []*strings.Builder{{}} - init := 4 - end := len(pattern) - 2 - // If the regex is wrapped in a group we can remove the first and last parentheses - if pattern[init] == '(' && pattern[end-1] == ')' { - init++ - end-- - } - for i := init; i < end; i++ { - if escaped { - switch { - case isRegexMetaCharacter(pattern[i]): - sets[len(sets)-1].WriteByte(pattern[i]) - case pattern[i] == '\\': - sets[len(sets)-1].WriteByte('\\') - default: - return nil - } - escaped = false - } else { - switch { - case isRegexMetaCharacter(pattern[i]): - if pattern[i] == '|' { - sets = append(sets, &strings.Builder{}) - } else { - return nil - } - case pattern[i] == '\\': - escaped = true - default: - sets[len(sets)-1].WriteByte(pattern[i]) - } - } - } - matches := make([]string, 0, len(sets)) - for _, s := range sets { - if s.Len() > 0 { - matches = append(matches, s.String()) - } - } - return matches -} - // PostingsForMatchers assembles a single postings iterator against the index reader // based on the given matchers. The resulting postings are not ordered by series. func PostingsForMatchers(ctx context.Context, ix IndexReader, ms ...*labels.Matcher) (index.Postings, error) { @@ -385,7 +320,7 @@ func postingsForMatcher(ctx context.Context, ix IndexReader, m *labels.Matcher) // Fast-path for set matching. if m.Type == labels.MatchRegexp { - setMatches := findSetMatches(m.GetRegexString()) + setMatches := m.SetMatches() if len(setMatches) > 0 { return ix.Postings(ctx, m.Name, setMatches...) } @@ -416,7 +351,7 @@ func inversePostingsForMatcher(ctx context.Context, ix IndexReader, m *labels.Ma // Inverse of a MatchNotRegexp is MatchRegexp (double negation). // Fast-path for set matching. if m.Type == labels.MatchNotRegexp { - setMatches := findSetMatches(m.GetRegexString()) + setMatches := m.SetMatches() if len(setMatches) > 0 { return ix.Postings(ctx, m.Name, setMatches...) } diff --git a/tsdb/querier_test.go b/tsdb/querier_test.go index 93704809b9..a293a983da 100644 --- a/tsdb/querier_test.go +++ b/tsdb/querier_test.go @@ -2658,54 +2658,6 @@ func BenchmarkSetMatcher(b *testing.B) { } } -// Refer to https://github.com/prometheus/prometheus/issues/2651. -func TestFindSetMatches(t *testing.T) { - cases := []struct { - pattern string - exp []string - }{ - // Single value, coming from a `bar=~"foo"` selector. - { - pattern: "^(?:foo)$", - exp: []string{ - "foo", - }, - }, - // Simple sets. - { - pattern: "^(?:foo|bar|baz)$", - exp: []string{ - "foo", - "bar", - "baz", - }, - }, - // Simple sets containing escaped characters. - { - pattern: "^(?:fo\\.o|bar\\?|\\^baz)$", - exp: []string{ - "fo.o", - "bar?", - "^baz", - }, - }, - // Simple sets containing special characters without escaping. - { - pattern: "^(?:fo.o|bar?|^baz)$", - exp: nil, - }, - // Missing wrapper. - { - pattern: "foo|bar|baz", - exp: nil, - }, - } - - for _, c := range cases { - require.Equal(t, c.exp, findSetMatches(c.pattern), "Evaluating %s, unexpected result.", c.pattern) - } -} - func TestPostingsForMatchers(t *testing.T) { ctx := context.Background() @@ -3310,7 +3262,7 @@ func TestPostingsForMatcher(t *testing.T) { { // Test case for double quoted regex matcher matcher: labels.MustNewMatcher(labels.MatchRegexp, "test", "^(?:a|b)$"), - hasError: true, + hasError: false, }, } diff --git a/tsdb/wal.go b/tsdb/wal.go deleted file mode 100644 index e06a8aea53..0000000000 --- a/tsdb/wal.go +++ /dev/null @@ -1,1303 +0,0 @@ -// Copyright 2017 The Prometheus Authors -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. - -package tsdb - -import ( - "bufio" - "encoding/binary" - "errors" - "fmt" - "hash" - "hash/crc32" - "io" - "math" - "os" - "path/filepath" - "sync" - "time" - - "github.com/go-kit/log" - "github.com/go-kit/log/level" - "github.com/prometheus/client_golang/prometheus" - - "github.com/prometheus/prometheus/model/labels" - "github.com/prometheus/prometheus/storage" - "github.com/prometheus/prometheus/tsdb/chunks" - "github.com/prometheus/prometheus/tsdb/encoding" - "github.com/prometheus/prometheus/tsdb/fileutil" - "github.com/prometheus/prometheus/tsdb/record" - "github.com/prometheus/prometheus/tsdb/tombstones" - "github.com/prometheus/prometheus/tsdb/wlog" - "github.com/prometheus/prometheus/util/zeropool" -) - -// WALEntryType indicates what data a WAL entry contains. -type WALEntryType uint8 - -const ( - // WALMagic is a 4 byte number every WAL segment file starts with. - WALMagic = uint32(0x43AF00EF) - - // WALFormatDefault is the version flag for the default outer segment file format. - WALFormatDefault = byte(1) -) - -// Entry types in a segment file. -const ( - WALEntrySymbols WALEntryType = 1 - WALEntrySeries WALEntryType = 2 - WALEntrySamples WALEntryType = 3 - WALEntryDeletes WALEntryType = 4 -) - -type walMetrics struct { - fsyncDuration prometheus.Summary - corruptions prometheus.Counter -} - -func newWalMetrics(r prometheus.Registerer) *walMetrics { - m := &walMetrics{} - - m.fsyncDuration = prometheus.NewSummary(prometheus.SummaryOpts{ - Name: "prometheus_tsdb_wal_fsync_duration_seconds", - Help: "Duration of WAL fsync.", - Objectives: map[float64]float64{0.5: 0.05, 0.9: 0.01, 0.99: 0.001}, - }) - m.corruptions = prometheus.NewCounter(prometheus.CounterOpts{ - Name: "prometheus_tsdb_wal_corruptions_total", - Help: "Total number of WAL corruptions.", - }) - - if r != nil { - r.MustRegister( - m.fsyncDuration, - m.corruptions, - ) - } - return m -} - -// WAL is a write ahead log that can log new series labels and samples. -// It must be completely read before new entries are logged. -// -// Deprecated: use wlog pkg combined with the record codex instead. -type WAL interface { - Reader() WALReader - LogSeries([]record.RefSeries) error - LogSamples([]record.RefSample) error - LogDeletes([]tombstones.Stone) error - Truncate(mint int64, keep func(uint64) bool) error - Close() error -} - -// WALReader reads entries from a WAL. -type WALReader interface { - Read( - seriesf func([]record.RefSeries), - samplesf func([]record.RefSample), - deletesf func([]tombstones.Stone), - ) error -} - -// segmentFile wraps a file object of a segment and tracks the highest timestamp -// it contains. During WAL truncating, all segments with no higher timestamp than -// the truncation threshold can be compacted. -type segmentFile struct { - *os.File - maxTime int64 // highest tombstone or sample timestamp in segment - minSeries chunks.HeadSeriesRef // lowerst series ID in segment -} - -func newSegmentFile(f *os.File) *segmentFile { - return &segmentFile{ - File: f, - maxTime: math.MinInt64, - minSeries: math.MaxUint64, - } -} - -const ( - walSegmentSizeBytes = 256 * 1024 * 1024 // 256 MB -) - -// The table gets initialized with sync.Once but may still cause a race -// with any other use of the crc32 package anywhere. Thus we initialize it -// before. -var castagnoliTable *crc32.Table - -func init() { - castagnoliTable = crc32.MakeTable(crc32.Castagnoli) -} - -// newCRC32 initializes a CRC32 hash with a preconfigured polynomial, so the -// polynomial may be easily changed in one location at a later time, if necessary. -func newCRC32() hash.Hash32 { - return crc32.New(castagnoliTable) -} - -// SegmentWAL is a write ahead log for series data. -// -// Deprecated: use wlog pkg combined with the record coders instead. -type SegmentWAL struct { - mtx sync.Mutex - metrics *walMetrics - - dirFile *os.File - files []*segmentFile - - logger log.Logger - flushInterval time.Duration - segmentSize int64 - - crc32 hash.Hash32 - cur *bufio.Writer - curN int64 - - stopc chan struct{} - donec chan struct{} - actorc chan func() error // sequentialized background operations - buffers sync.Pool -} - -// OpenSegmentWAL opens or creates a write ahead log in the given directory. -// The WAL must be read completely before new data is written. -func OpenSegmentWAL(dir string, logger log.Logger, flushInterval time.Duration, r prometheus.Registerer) (*SegmentWAL, error) { - if err := os.MkdirAll(dir, 0o777); err != nil { - return nil, err - } - df, err := fileutil.OpenDir(dir) - if err != nil { - return nil, err - } - if logger == nil { - logger = log.NewNopLogger() - } - - w := &SegmentWAL{ - dirFile: df, - logger: logger, - flushInterval: flushInterval, - donec: make(chan struct{}), - stopc: make(chan struct{}), - actorc: make(chan func() error, 2), - segmentSize: walSegmentSizeBytes, - crc32: newCRC32(), - } - w.metrics = newWalMetrics(r) - - fns, err := sequenceFiles(w.dirFile.Name()) - if err != nil { - return nil, err - } - - for i, fn := range fns { - f, err := w.openSegmentFile(fn) - if err == nil { - w.files = append(w.files, newSegmentFile(f)) - continue - } - level.Warn(logger).Log("msg", "Invalid segment file detected, truncating WAL", "err", err, "file", fn) - - for _, fn := range fns[i:] { - if err := os.Remove(fn); err != nil { - return w, fmt.Errorf("removing segment failed: %w", err) - } - } - break - } - - go w.run(flushInterval) - - return w, nil -} - -// repairingWALReader wraps a WAL reader and truncates its underlying SegmentWAL after the last -// valid entry if it encounters corruption. -type repairingWALReader struct { - wal *SegmentWAL - r WALReader -} - -func (r *repairingWALReader) Read( - seriesf func([]record.RefSeries), - samplesf func([]record.RefSample), - deletesf func([]tombstones.Stone), -) error { - err := r.r.Read(seriesf, samplesf, deletesf) - if err == nil { - return nil - } - var cerr *walCorruptionErr - if !errors.As(err, &cerr) { - return err - } - r.wal.metrics.corruptions.Inc() - return r.wal.truncate(cerr.err, cerr.file, cerr.lastOffset) -} - -// truncate the WAL after the last valid entry. -func (w *SegmentWAL) truncate(err error, file int, lastOffset int64) error { - level.Error(w.logger).Log("msg", "WAL corruption detected; truncating", - "err", err, "file", w.files[file].Name(), "pos", lastOffset) - - // Close and delete all files after the current one. - for _, f := range w.files[file+1:] { - if err := f.Close(); err != nil { - return err - } - if err := os.Remove(f.Name()); err != nil { - return err - } - } - w.mtx.Lock() - defer w.mtx.Unlock() - - w.files = w.files[:file+1] - - // Seek the current file to the last valid offset where we continue writing from. - _, err = w.files[file].Seek(lastOffset, io.SeekStart) - return err -} - -// Reader returns a new reader over the write ahead log data. -// It must be completely consumed before writing to the WAL. -func (w *SegmentWAL) Reader() WALReader { - return &repairingWALReader{ - wal: w, - r: newWALReader(w.files, w.logger), - } -} - -func (w *SegmentWAL) getBuffer() *encoding.Encbuf { - b := w.buffers.Get() - if b == nil { - return &encoding.Encbuf{B: make([]byte, 0, 64*1024)} - } - return b.(*encoding.Encbuf) -} - -func (w *SegmentWAL) putBuffer(b *encoding.Encbuf) { - b.Reset() - w.buffers.Put(b) -} - -// Truncate deletes the values prior to mint and the series which the keep function -// does not indicate to preserve. -func (w *SegmentWAL) Truncate(mint int64, keep func(chunks.HeadSeriesRef) bool) error { - // The last segment is always active. - if len(w.files) < 2 { - return nil - } - var candidates []*segmentFile - - // All files have to be traversed as there could be two segments for a block - // with first block having times (10000, 20000) and SECOND one having (0, 10000). - for _, sf := range w.files[:len(w.files)-1] { - if sf.maxTime >= mint { - break - } - // Past WAL files are closed. We have to reopen them for another read. - f, err := w.openSegmentFile(sf.Name()) - if err != nil { - return fmt.Errorf("open old WAL segment for read: %w", err) - } - candidates = append(candidates, &segmentFile{ - File: f, - minSeries: sf.minSeries, - maxTime: sf.maxTime, - }) - } - if len(candidates) == 0 { - return nil - } - - r := newWALReader(candidates, w.logger) - - // Create a new tmp file. - f, err := w.createSegmentFile(filepath.Join(w.dirFile.Name(), "compact.tmp")) - if err != nil { - return fmt.Errorf("create compaction segment: %w", err) - } - defer func() { - if err := os.RemoveAll(f.Name()); err != nil { - level.Error(w.logger).Log("msg", "remove tmp file", "err", err.Error()) - } - }() - - var ( - csf = newSegmentFile(f) - crc32 = newCRC32() - decSeries = []record.RefSeries{} - activeSeries = []record.RefSeries{} - ) - - for r.next() { - rt, flag, byt := r.at() - - if rt != WALEntrySeries { - continue - } - decSeries = decSeries[:0] - activeSeries = activeSeries[:0] - - err := r.decodeSeries(flag, byt, &decSeries) - if err != nil { - return fmt.Errorf("decode samples while truncating: %w", err) - } - for _, s := range decSeries { - if keep(s.Ref) { - activeSeries = append(activeSeries, s) - } - } - - buf := w.getBuffer() - flag = w.encodeSeries(buf, activeSeries) - - _, err = w.writeTo(csf, crc32, WALEntrySeries, flag, buf.Get()) - w.putBuffer(buf) - - if err != nil { - return fmt.Errorf("write to compaction segment: %w", err) - } - } - if err := r.Err(); err != nil { - return fmt.Errorf("read candidate WAL files: %w", err) - } - - off, err := csf.Seek(0, io.SeekCurrent) - if err != nil { - return err - } - if err := csf.Truncate(off); err != nil { - return err - } - if err := csf.Sync(); err != nil { - return nil - } - if err := csf.Close(); err != nil { - return nil - } - - _ = candidates[0].Close() // need close before remove on platform windows - if err := fileutil.Replace(csf.Name(), candidates[0].Name()); err != nil { - return fmt.Errorf("rename compaction segment: %w", err) - } - for _, f := range candidates[1:] { - f.Close() // need close before remove on platform windows - if err := os.RemoveAll(f.Name()); err != nil { - return fmt.Errorf("delete WAL segment file: %w", err) - } - } - if err := w.dirFile.Sync(); err != nil { - return err - } - - // The file object of csf still holds the name before rename. Recreate it so - // subsequent truncations do not look at a non-existent file name. - csf.File, err = w.openSegmentFile(candidates[0].Name()) - if err != nil { - return err - } - // We don't need it to be open. - if err := csf.Close(); err != nil { - return err - } - - w.mtx.Lock() - w.files = append([]*segmentFile{csf}, w.files[len(candidates):]...) - w.mtx.Unlock() - - return nil -} - -// LogSeries writes a batch of new series labels to the log. -// The series have to be ordered. -func (w *SegmentWAL) LogSeries(series []record.RefSeries) error { - buf := w.getBuffer() - - flag := w.encodeSeries(buf, series) - - w.mtx.Lock() - defer w.mtx.Unlock() - - err := w.write(WALEntrySeries, flag, buf.Get()) - - w.putBuffer(buf) - - if err != nil { - return fmt.Errorf("log series: %w", err) - } - - tf := w.head() - - for _, s := range series { - if tf.minSeries > s.Ref { - tf.minSeries = s.Ref - } - } - return nil -} - -// LogSamples writes a batch of new samples to the log. -func (w *SegmentWAL) LogSamples(samples []record.RefSample) error { - buf := w.getBuffer() - - flag := w.encodeSamples(buf, samples) - - w.mtx.Lock() - defer w.mtx.Unlock() - - err := w.write(WALEntrySamples, flag, buf.Get()) - - w.putBuffer(buf) - - if err != nil { - return fmt.Errorf("log series: %w", err) - } - tf := w.head() - - for _, s := range samples { - if tf.maxTime < s.T { - tf.maxTime = s.T - } - } - return nil -} - -// LogDeletes write a batch of new deletes to the log. -func (w *SegmentWAL) LogDeletes(stones []tombstones.Stone) error { - buf := w.getBuffer() - - flag := w.encodeDeletes(buf, stones) - - w.mtx.Lock() - defer w.mtx.Unlock() - - err := w.write(WALEntryDeletes, flag, buf.Get()) - - w.putBuffer(buf) - - if err != nil { - return fmt.Errorf("log series: %w", err) - } - tf := w.head() - - for _, s := range stones { - for _, iv := range s.Intervals { - if tf.maxTime < iv.Maxt { - tf.maxTime = iv.Maxt - } - } - } - return nil -} - -// openSegmentFile opens the given segment file and consumes and validates header. -func (w *SegmentWAL) openSegmentFile(name string) (*os.File, error) { - // We must open all files in read/write mode as we may have to truncate along - // the way and any file may become the head. - f, err := os.OpenFile(name, os.O_RDWR, 0o666) - if err != nil { - return nil, err - } - metab := make([]byte, 8) - - // If there is an error, we need close f for platform windows before gc. - // Otherwise, file op may fail. - hasError := true - defer func() { - if hasError { - f.Close() - } - }() - - switch n, err := f.Read(metab); { - case err != nil: - return nil, fmt.Errorf("validate meta %q: %w", f.Name(), err) - case n != 8: - return nil, fmt.Errorf("invalid header size %d in %q", n, f.Name()) - } - - if m := binary.BigEndian.Uint32(metab[:4]); m != WALMagic { - return nil, fmt.Errorf("invalid magic header %x in %q", m, f.Name()) - } - if metab[4] != WALFormatDefault { - return nil, fmt.Errorf("unknown WAL segment format %d in %q", metab[4], f.Name()) - } - hasError = false - return f, nil -} - -// createSegmentFile creates a new segment file with the given name. It preallocates -// the standard segment size if possible and writes the header. -func (w *SegmentWAL) createSegmentFile(name string) (*os.File, error) { - f, err := os.Create(name) - if err != nil { - return nil, err - } - if err = fileutil.Preallocate(f, w.segmentSize, true); err != nil { - return nil, err - } - // Write header metadata for new file. - metab := make([]byte, 8) - binary.BigEndian.PutUint32(metab[:4], WALMagic) - metab[4] = WALFormatDefault - - if _, err := f.Write(metab); err != nil { - return nil, err - } - return f, err -} - -// cut finishes the currently active segments and opens the next one. -// The encoder is reset to point to the new segment. -func (w *SegmentWAL) cut() error { - // Sync current head to disk and close. - if hf := w.head(); hf != nil { - if err := w.flush(); err != nil { - return err - } - // Finish last segment asynchronously to not block the WAL moving along - // in the new segment. - go func() { - w.actorc <- func() error { - off, err := hf.Seek(0, io.SeekCurrent) - if err != nil { - return fmt.Errorf("finish old segment %s: %w", hf.Name(), err) - } - if err := hf.Truncate(off); err != nil { - return fmt.Errorf("finish old segment %s: %w", hf.Name(), err) - } - if err := hf.Sync(); err != nil { - return fmt.Errorf("finish old segment %s: %w", hf.Name(), err) - } - if err := hf.Close(); err != nil { - return fmt.Errorf("finish old segment %s: %w", hf.Name(), err) - } - return nil - } - }() - } - - p, _, err := nextSequenceFile(w.dirFile.Name()) - if err != nil { - return err - } - f, err := w.createSegmentFile(p) - if err != nil { - return err - } - - go func() { - w.actorc <- func() error { - if err := w.dirFile.Sync(); err != nil { - return fmt.Errorf("sync WAL directory: %w", err) - } - return nil - } - }() - - w.files = append(w.files, newSegmentFile(f)) - - // TODO(gouthamve): make the buffer size a constant. - w.cur = bufio.NewWriterSize(f, 8*1024*1024) - w.curN = 8 - - return nil -} - -func (w *SegmentWAL) head() *segmentFile { - if len(w.files) == 0 { - return nil - } - return w.files[len(w.files)-1] -} - -// Sync flushes the changes to disk. -func (w *SegmentWAL) Sync() error { - var head *segmentFile - var err error - - // Flush the writer and retrieve the reference to the head segment under mutex lock. - func() { - w.mtx.Lock() - defer w.mtx.Unlock() - if err = w.flush(); err != nil { - return - } - head = w.head() - }() - if err != nil { - return fmt.Errorf("flush buffer: %w", err) - } - if head != nil { - // But only fsync the head segment after releasing the mutex as it will block on disk I/O. - start := time.Now() - err := fileutil.Fdatasync(head.File) - w.metrics.fsyncDuration.Observe(time.Since(start).Seconds()) - return err - } - return nil -} - -func (w *SegmentWAL) sync() error { - if err := w.flush(); err != nil { - return err - } - if w.head() == nil { - return nil - } - - start := time.Now() - err := fileutil.Fdatasync(w.head().File) - w.metrics.fsyncDuration.Observe(time.Since(start).Seconds()) - return err -} - -func (w *SegmentWAL) flush() error { - if w.cur == nil { - return nil - } - return w.cur.Flush() -} - -func (w *SegmentWAL) run(interval time.Duration) { - var tick <-chan time.Time - - if interval > 0 { - ticker := time.NewTicker(interval) - defer ticker.Stop() - tick = ticker.C - } - defer close(w.donec) - - for { - // Processing all enqueued operations has precedence over shutdown and - // background syncs. - select { - case f := <-w.actorc: - if err := f(); err != nil { - level.Error(w.logger).Log("msg", "operation failed", "err", err) - } - continue - default: - } - select { - case <-w.stopc: - return - case f := <-w.actorc: - if err := f(); err != nil { - level.Error(w.logger).Log("msg", "operation failed", "err", err) - } - case <-tick: - if err := w.Sync(); err != nil { - level.Error(w.logger).Log("msg", "sync failed", "err", err) - } - } - } -} - -// Close syncs all data and closes the underlying resources. -func (w *SegmentWAL) Close() error { - // Make sure you can call Close() multiple times. - select { - case <-w.stopc: - return nil // Already closed. - default: - } - - close(w.stopc) - <-w.donec - - w.mtx.Lock() - defer w.mtx.Unlock() - - if err := w.sync(); err != nil { - return err - } - // On opening, a WAL must be fully consumed once. Afterwards - // only the current segment will still be open. - if hf := w.head(); hf != nil { - if err := hf.Close(); err != nil { - return fmt.Errorf("closing WAL head %s: %w", hf.Name(), err) - } - } - if err := w.dirFile.Close(); err != nil { - return fmt.Errorf("closing WAL dir %s: %w", w.dirFile.Name(), err) - } - return nil -} - -func (w *SegmentWAL) write(t WALEntryType, flag uint8, buf []byte) error { - // Cut to the next segment if the entry exceeds the file size unless it would also - // exceed the size of a new segment. - // TODO(gouthamve): Add a test for this case where the commit is greater than segmentSize. - var ( - sz = int64(len(buf)) + 6 - newsz = w.curN + sz - ) - // XXX(fabxc): this currently cuts a new file whenever the WAL was newly opened. - // Probably fine in general but may yield a lot of short files in some cases. - if w.cur == nil || w.curN > w.segmentSize || newsz > w.segmentSize && sz <= w.segmentSize { - if err := w.cut(); err != nil { - return err - } - } - n, err := w.writeTo(w.cur, w.crc32, t, flag, buf) - - w.curN += int64(n) - - return err -} - -func (w *SegmentWAL) writeTo(wr io.Writer, crc32 hash.Hash, t WALEntryType, flag uint8, buf []byte) (int, error) { - if len(buf) == 0 { - return 0, nil - } - crc32.Reset() - wr = io.MultiWriter(crc32, wr) - - var b [6]byte - b[0] = byte(t) - b[1] = flag - - binary.BigEndian.PutUint32(b[2:], uint32(len(buf))) - - n1, err := wr.Write(b[:]) - if err != nil { - return n1, err - } - n2, err := wr.Write(buf) - if err != nil { - return n1 + n2, err - } - n3, err := wr.Write(crc32.Sum(b[:0])) - - return n1 + n2 + n3, err -} - -const ( - walSeriesSimple = 1 - walSamplesSimple = 1 - walDeletesSimple = 1 -) - -func (w *SegmentWAL) encodeSeries(buf *encoding.Encbuf, series []record.RefSeries) uint8 { - for _, s := range series { - buf.PutBE64(uint64(s.Ref)) - record.EncodeLabels(buf, s.Labels) - } - return walSeriesSimple -} - -func (w *SegmentWAL) encodeSamples(buf *encoding.Encbuf, samples []record.RefSample) uint8 { - if len(samples) == 0 { - return walSamplesSimple - } - // Store base timestamp and base reference number of first sample. - // All samples encode their timestamp and ref as delta to those. - // - // TODO(fabxc): optimize for all samples having the same timestamp. - first := samples[0] - - buf.PutBE64(uint64(first.Ref)) - buf.PutBE64int64(first.T) - - for _, s := range samples { - buf.PutVarint64(int64(s.Ref) - int64(first.Ref)) - buf.PutVarint64(s.T - first.T) - buf.PutBE64(math.Float64bits(s.V)) - } - return walSamplesSimple -} - -func (w *SegmentWAL) encodeDeletes(buf *encoding.Encbuf, stones []tombstones.Stone) uint8 { - for _, s := range stones { - for _, iv := range s.Intervals { - buf.PutBE64(uint64(s.Ref)) - buf.PutVarint64(iv.Mint) - buf.PutVarint64(iv.Maxt) - } - } - return walDeletesSimple -} - -// walReader decodes and emits write ahead log entries. -type walReader struct { - logger log.Logger - - files []*segmentFile - cur int - buf []byte - crc32 hash.Hash32 - dec record.Decoder - - curType WALEntryType - curFlag byte - curBuf []byte - lastOffset int64 // offset after last successfully read entry - - err error -} - -func newWALReader(files []*segmentFile, l log.Logger) *walReader { - if l == nil { - l = log.NewNopLogger() - } - return &walReader{ - logger: l, - files: files, - buf: make([]byte, 0, 128*4096), - crc32: newCRC32(), - dec: record.NewDecoder(labels.NewSymbolTable()), - } -} - -// Err returns the last error the reader encountered. -func (r *walReader) Err() error { - return r.err -} - -func (r *walReader) Read( - seriesf func([]record.RefSeries), - samplesf func([]record.RefSample), - deletesf func([]tombstones.Stone), -) error { - // Concurrency for replaying the WAL is very limited. We at least split out decoding and - // processing into separate threads. - // Historically, the processing is the bottleneck with reading and decoding using only - // 15% of the CPU. - var ( - seriesPool zeropool.Pool[[]record.RefSeries] - samplePool zeropool.Pool[[]record.RefSample] - deletePool zeropool.Pool[[]tombstones.Stone] - ) - donec := make(chan struct{}) - datac := make(chan interface{}, 100) - - go func() { - defer close(donec) - - for x := range datac { - switch v := x.(type) { - case []record.RefSeries: - if seriesf != nil { - seriesf(v) - } - seriesPool.Put(v[:0]) - case []record.RefSample: - if samplesf != nil { - samplesf(v) - } - samplePool.Put(v[:0]) - case []tombstones.Stone: - if deletesf != nil { - deletesf(v) - } - deletePool.Put(v[:0]) - default: - level.Error(r.logger).Log("msg", "unexpected data type") - } - } - }() - - var err error - - for r.next() { - et, flag, b := r.at() - - // In decoding below we never return a walCorruptionErr for now. - // Those should generally be caught by entry decoding before. - switch et { - case WALEntrySeries: - series := seriesPool.Get() - if series == nil { - series = make([]record.RefSeries, 0, 512) - } - - err = r.decodeSeries(flag, b, &series) - if err != nil { - err = fmt.Errorf("decode series entry: %w", err) - break - } - datac <- series - - cf := r.current() - for _, s := range series { - if cf.minSeries > s.Ref { - cf.minSeries = s.Ref - } - } - case WALEntrySamples: - samples := samplePool.Get() - if samples == nil { - samples = make([]record.RefSample, 0, 512) - } - - err = r.decodeSamples(flag, b, &samples) - if err != nil { - err = fmt.Errorf("decode samples entry: %w", err) - break - } - datac <- samples - - // Update the times for the WAL segment file. - cf := r.current() - for _, s := range samples { - if cf.maxTime < s.T { - cf.maxTime = s.T - } - } - case WALEntryDeletes: - deletes := deletePool.Get() - if deletes == nil { - deletes = make([]tombstones.Stone, 0, 512) - } - - err = r.decodeDeletes(flag, b, &deletes) - if err != nil { - err = fmt.Errorf("decode delete entry: %w", err) - break - } - datac <- deletes - - // Update the times for the WAL segment file. - cf := r.current() - for _, s := range deletes { - for _, iv := range s.Intervals { - if cf.maxTime < iv.Maxt { - cf.maxTime = iv.Maxt - } - } - } - } - } - close(datac) - <-donec - - if err != nil { - return err - } - if err := r.Err(); err != nil { - return fmt.Errorf("read entry: %w", err) - } - return nil -} - -func (r *walReader) at() (WALEntryType, byte, []byte) { - return r.curType, r.curFlag, r.curBuf -} - -// next returns decodes the next entry pair and returns true -// if it was successful. -func (r *walReader) next() bool { - if r.cur >= len(r.files) { - return false - } - cf := r.files[r.cur] - - // Remember the offset after the last correctly read entry. If the next one - // is corrupted, this is where we can safely truncate. - r.lastOffset, r.err = cf.Seek(0, io.SeekCurrent) - if r.err != nil { - return false - } - - et, flag, b, err := r.entry(cf) - // If we reached the end of the reader, advance to the next one - // and close. - // Do not close on the last one as it will still be appended to. - if errors.Is(err, io.EOF) { - if r.cur == len(r.files)-1 { - return false - } - // Current reader completed, close and move to the next one. - if err := cf.Close(); err != nil { - r.err = err - return false - } - r.cur++ - return r.next() - } - if err != nil { - r.err = err - return false - } - - r.curType = et - r.curFlag = flag - r.curBuf = b - return r.err == nil -} - -func (r *walReader) current() *segmentFile { - return r.files[r.cur] -} - -// walCorruptionErr is a type wrapper for errors that indicate WAL corruption -// and trigger a truncation. -type walCorruptionErr struct { - err error - file int - lastOffset int64 -} - -func (e *walCorruptionErr) Error() string { - return fmt.Sprintf("%s ", e.err, e.file, e.lastOffset) -} - -func (e *walCorruptionErr) Unwrap() error { - return e.err -} - -func (r *walReader) corruptionErr(s string, args ...interface{}) error { - return &walCorruptionErr{ - err: fmt.Errorf(s, args...), - file: r.cur, - lastOffset: r.lastOffset, - } -} - -func (r *walReader) entry(cr io.Reader) (WALEntryType, byte, []byte, error) { - r.crc32.Reset() - tr := io.TeeReader(cr, r.crc32) - - b := make([]byte, 6) - switch n, err := tr.Read(b); { - case err != nil: - return 0, 0, nil, err - case n != 6: - return 0, 0, nil, r.corruptionErr("invalid entry header size %d", n) - } - - var ( - etype = WALEntryType(b[0]) - flag = b[1] - length = int(binary.BigEndian.Uint32(b[2:])) - ) - // Exit if we reached pre-allocated space. - if etype == 0 { - return 0, 0, nil, io.EOF - } - if etype != WALEntrySeries && etype != WALEntrySamples && etype != WALEntryDeletes { - return 0, 0, nil, r.corruptionErr("invalid entry type %d", etype) - } - - if length > len(r.buf) { - r.buf = make([]byte, length) - } - buf := r.buf[:length] - - switch n, err := tr.Read(buf); { - case err != nil: - return 0, 0, nil, err - case n != length: - return 0, 0, nil, r.corruptionErr("invalid entry body size %d", n) - } - - switch n, err := cr.Read(b[:4]); { - case err != nil: - return 0, 0, nil, err - case n != 4: - return 0, 0, nil, r.corruptionErr("invalid checksum length %d", n) - } - if exp, has := binary.BigEndian.Uint32(b[:4]), r.crc32.Sum32(); has != exp { - return 0, 0, nil, r.corruptionErr("unexpected CRC32 checksum %x, want %x", has, exp) - } - - return etype, flag, buf, nil -} - -func (r *walReader) decodeSeries(flag byte, b []byte, res *[]record.RefSeries) error { - dec := encoding.Decbuf{B: b} - - for len(dec.B) > 0 && dec.Err() == nil { - ref := chunks.HeadSeriesRef(dec.Be64()) - lset := r.dec.DecodeLabels(&dec) - - *res = append(*res, record.RefSeries{ - Ref: ref, - Labels: lset, - }) - } - if dec.Err() != nil { - return dec.Err() - } - if len(dec.B) > 0 { - return fmt.Errorf("unexpected %d bytes left in entry", len(dec.B)) - } - return nil -} - -func (r *walReader) decodeSamples(flag byte, b []byte, res *[]record.RefSample) error { - if len(b) == 0 { - return nil - } - dec := encoding.Decbuf{B: b} - - var ( - baseRef = dec.Be64() - baseTime = dec.Be64int64() - ) - - for len(dec.B) > 0 && dec.Err() == nil { - dref := dec.Varint64() - dtime := dec.Varint64() - val := dec.Be64() - - *res = append(*res, record.RefSample{ - Ref: chunks.HeadSeriesRef(int64(baseRef) + dref), - T: baseTime + dtime, - V: math.Float64frombits(val), - }) - } - - if err := dec.Err(); err != nil { - return fmt.Errorf("decode error after %d samples: %w", len(*res), err) - } - if len(dec.B) > 0 { - return fmt.Errorf("unexpected %d bytes left in entry", len(dec.B)) - } - return nil -} - -func (r *walReader) decodeDeletes(flag byte, b []byte, res *[]tombstones.Stone) error { - dec := &encoding.Decbuf{B: b} - - for dec.Len() > 0 && dec.Err() == nil { - *res = append(*res, tombstones.Stone{ - Ref: storage.SeriesRef(dec.Be64()), - Intervals: tombstones.Intervals{ - {Mint: dec.Varint64(), Maxt: dec.Varint64()}, - }, - }) - } - if dec.Err() != nil { - return dec.Err() - } - if len(dec.B) > 0 { - return fmt.Errorf("unexpected %d bytes left in entry", len(dec.B)) - } - return nil -} - -func deprecatedWALExists(logger log.Logger, dir string) (bool, error) { - // Detect whether we still have the old WAL. - fns, err := sequenceFiles(dir) - if err != nil && !os.IsNotExist(err) { - return false, fmt.Errorf("list sequence files: %w", err) - } - if len(fns) == 0 { - return false, nil // No WAL at all yet. - } - // Check header of first segment to see whether we are still dealing with an - // old WAL. - f, err := os.Open(fns[0]) - if err != nil { - return false, fmt.Errorf("check first existing segment: %w", err) - } - defer f.Close() - - var hdr [4]byte - if _, err := f.Read(hdr[:]); err != nil && !errors.Is(err, io.EOF) { - return false, fmt.Errorf("read header from first segment: %w", err) - } - // If we cannot read the magic header for segments of the old WAL, abort. - // Either it's migrated already or there's a corruption issue with which - // we cannot deal here anyway. Subsequent attempts to open the WAL will error in that case. - if binary.BigEndian.Uint32(hdr[:]) != WALMagic { - return false, nil - } - return true, nil -} - -// MigrateWAL rewrites the deprecated write ahead log into the new format. -func MigrateWAL(logger log.Logger, dir string) (err error) { - if logger == nil { - logger = log.NewNopLogger() - } - if exists, err := deprecatedWALExists(logger, dir); err != nil || !exists { - return err - } - level.Info(logger).Log("msg", "Migrating WAL format") - - tmpdir := dir + ".tmp" - if err := os.RemoveAll(tmpdir); err != nil { - return fmt.Errorf("cleanup replacement dir: %w", err) - } - repl, err := wlog.New(logger, nil, tmpdir, wlog.CompressionNone) - if err != nil { - return fmt.Errorf("open new WAL: %w", err) - } - - // It should've already been closed as part of the previous finalization. - // Do it once again in case of prior errors. - defer func() { - if err != nil { - repl.Close() - } - }() - - w, err := OpenSegmentWAL(dir, logger, time.Minute, nil) - if err != nil { - return fmt.Errorf("open old WAL: %w", err) - } - defer w.Close() - - rdr := w.Reader() - - var ( - enc record.Encoder - b []byte - ) - decErr := rdr.Read( - func(s []record.RefSeries) { - if err != nil { - return - } - err = repl.Log(enc.Series(s, b[:0])) - }, - func(s []record.RefSample) { - if err != nil { - return - } - err = repl.Log(enc.Samples(s, b[:0])) - }, - func(s []tombstones.Stone) { - if err != nil { - return - } - err = repl.Log(enc.Tombstones(s, b[:0])) - }, - ) - if decErr != nil { - return fmt.Errorf("decode old entries: %w", err) - } - if err != nil { - return fmt.Errorf("write new entries: %w", err) - } - // We explicitly close even when there is a defer for Windows to be - // able to delete it. The defer is in place to close it in-case there - // are errors above. - if err := w.Close(); err != nil { - return fmt.Errorf("close old WAL: %w", err) - } - if err := repl.Close(); err != nil { - return fmt.Errorf("close new WAL: %w", err) - } - if err := fileutil.Replace(tmpdir, dir); err != nil { - return fmt.Errorf("replace old WAL: %w", err) - } - return nil -} diff --git a/tsdb/wal_test.go b/tsdb/wal_test.go deleted file mode 100644 index 7794a54547..0000000000 --- a/tsdb/wal_test.go +++ /dev/null @@ -1,553 +0,0 @@ -// Copyright 2017 The Prometheus Authors -// Licensed under the Apache License, Version 2.0 (the "License"); -// you may not use this file except in compliance with the License. -// You may obtain a copy of the License at -// -// http://www.apache.org/licenses/LICENSE-2.0 -// -// Unless required by applicable law or agreed to in writing, software -// distributed under the License is distributed on an "AS IS" BASIS, -// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. -// See the License for the specific language governing permissions and -// limitations under the License. - -//go:build !windows - -package tsdb - -import ( - "encoding/binary" - "io" - "math/rand" - "os" - "path" - "path/filepath" - "testing" - "time" - - "github.com/go-kit/log" - "github.com/stretchr/testify/require" - - "github.com/prometheus/prometheus/model/labels" - "github.com/prometheus/prometheus/storage" - "github.com/prometheus/prometheus/tsdb/chunks" - "github.com/prometheus/prometheus/tsdb/record" - "github.com/prometheus/prometheus/tsdb/tombstones" - "github.com/prometheus/prometheus/tsdb/wlog" - "github.com/prometheus/prometheus/util/testutil" -) - -func TestSegmentWAL_cut(t *testing.T) { - tmpdir := t.TempDir() - - // This calls cut() implicitly the first time without a previous tail. - w, err := OpenSegmentWAL(tmpdir, nil, 0, nil) - require.NoError(t, err) - - require.NoError(t, w.write(WALEntrySeries, 1, []byte("Hello World!!"))) - - require.NoError(t, w.cut()) - - // Cutting creates a new file. - require.Len(t, w.files, 2) - - require.NoError(t, w.write(WALEntrySeries, 1, []byte("Hello World!!"))) - - require.NoError(t, w.Close()) - - for _, of := range w.files { - f, err := os.Open(of.Name()) - require.NoError(t, err) - - // Verify header data. - metab := make([]byte, 8) - _, err = f.Read(metab) - require.NoError(t, err) - require.Equal(t, WALMagic, binary.BigEndian.Uint32(metab[:4])) - require.Equal(t, WALFormatDefault, metab[4]) - - // We cannot actually check for correct pre-allocation as it is - // optional per filesystem and handled transparently. - et, flag, b, err := newWALReader(nil, nil).entry(f) - require.NoError(t, err) - require.Equal(t, WALEntrySeries, et) - require.Equal(t, byte(walSeriesSimple), flag) - require.Equal(t, []byte("Hello World!!"), b) - } -} - -func TestSegmentWAL_Truncate(t *testing.T) { - const ( - numMetrics = 20000 - batch = 100 - ) - series, err := labels.ReadLabels(filepath.Join("testdata", "20kseries.json"), numMetrics) - require.NoError(t, err) - - dir := t.TempDir() - - w, err := OpenSegmentWAL(dir, nil, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(w) - w.segmentSize = 10000 - - for i := 0; i < numMetrics; i += batch { - var rs []record.RefSeries - - for j, s := range series[i : i+batch] { - rs = append(rs, record.RefSeries{Labels: s, Ref: chunks.HeadSeriesRef(i+j) + 1}) - } - err := w.LogSeries(rs) - require.NoError(t, err) - } - - // We mark the 2nd half of the files with a min timestamp that should discard - // them from the selection of compactable files. - for i, f := range w.files[len(w.files)/2:] { - f.maxTime = int64(1000 + i) - } - // All series in those files must be preserved regarding of the provided postings list. - boundarySeries := w.files[len(w.files)/2].minSeries - - // We truncate while keeping every 2nd series. - keep := map[chunks.HeadSeriesRef]struct{}{} - for i := 1; i <= numMetrics; i += 2 { - keep[chunks.HeadSeriesRef(i)] = struct{}{} - } - keepf := func(id chunks.HeadSeriesRef) bool { - _, ok := keep[id] - return ok - } - - err = w.Truncate(1000, keepf) - require.NoError(t, err) - - var expected []record.RefSeries - - for i := 1; i <= numMetrics; i++ { - if i%2 == 1 || chunks.HeadSeriesRef(i) >= boundarySeries { - expected = append(expected, record.RefSeries{Ref: chunks.HeadSeriesRef(i), Labels: series[i-1]}) - } - } - - // Call Truncate once again to see whether we can read the written file without - // creating a new WAL. - err = w.Truncate(1000, keepf) - require.NoError(t, err) - require.NoError(t, w.Close()) - - // The same again with a new WAL. - w, err = OpenSegmentWAL(dir, nil, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(w) - - var readSeries []record.RefSeries - r := w.Reader() - - require.NoError(t, r.Read(func(s []record.RefSeries) { - readSeries = append(readSeries, s...) - }, nil, nil)) - - testutil.RequireEqual(t, expected, readSeries) -} - -// Symmetrical test of reading and writing to the WAL via its main interface. -func TestSegmentWAL_Log_Restore(t *testing.T) { - const ( - numMetrics = 50 - iterations = 5 - stepSize = 5 - ) - // Generate testing data. It does not make semantic sense but - // for the purpose of this test. - series, err := labels.ReadLabels(filepath.Join("testdata", "20kseries.json"), numMetrics) - require.NoError(t, err) - - dir := t.TempDir() - - var ( - recordedSeries [][]record.RefSeries - recordedSamples [][]record.RefSample - recordedDeletes [][]tombstones.Stone - ) - var totalSamples int - - // Open WAL a bunch of times, validate all previous data can be read, - // write more data to it, close it. - for k := 0; k < numMetrics; k += numMetrics / iterations { - w, err := OpenSegmentWAL(dir, nil, 0, nil) - require.NoError(t, err) - - // Set smaller segment size so we can actually write several files. - w.segmentSize = 1000 * 1000 - - r := w.Reader() - - var ( - resultSeries [][]record.RefSeries - resultSamples [][]record.RefSample - resultDeletes [][]tombstones.Stone - ) - - serf := func(series []record.RefSeries) { - if len(series) > 0 { - clsets := make([]record.RefSeries, len(series)) - copy(clsets, series) - resultSeries = append(resultSeries, clsets) - } - } - smplf := func(smpls []record.RefSample) { - if len(smpls) > 0 { - csmpls := make([]record.RefSample, len(smpls)) - copy(csmpls, smpls) - resultSamples = append(resultSamples, csmpls) - } - } - - delf := func(stones []tombstones.Stone) { - if len(stones) > 0 { - cst := make([]tombstones.Stone, len(stones)) - copy(cst, stones) - resultDeletes = append(resultDeletes, cst) - } - } - - require.NoError(t, r.Read(serf, smplf, delf)) - - testutil.RequireEqual(t, recordedSamples, resultSamples) - testutil.RequireEqual(t, recordedSeries, resultSeries) - testutil.RequireEqual(t, recordedDeletes, resultDeletes) - - series := series[k : k+(numMetrics/iterations)] - - // Insert in batches and generate different amounts of samples for each. - for i := 0; i < len(series); i += stepSize { - var samples []record.RefSample - var stones []tombstones.Stone - - for j := 0; j < i*10; j++ { - samples = append(samples, record.RefSample{ - Ref: chunks.HeadSeriesRef(j % 10000), - T: int64(j * 2), - V: rand.Float64(), - }) - } - - for j := 0; j < i*20; j++ { - ts := rand.Int63() - stones = append(stones, tombstones.Stone{Ref: storage.SeriesRef(rand.Uint64()), Intervals: tombstones.Intervals{{Mint: ts, Maxt: ts + rand.Int63n(10000)}}}) - } - - lbls := series[i : i+stepSize] - series := make([]record.RefSeries, 0, len(series)) - for j, l := range lbls { - series = append(series, record.RefSeries{ - Ref: chunks.HeadSeriesRef(i + j), - Labels: l, - }) - } - - require.NoError(t, w.LogSeries(series)) - require.NoError(t, w.LogSamples(samples)) - require.NoError(t, w.LogDeletes(stones)) - - if len(lbls) > 0 { - recordedSeries = append(recordedSeries, series) - } - if len(samples) > 0 { - recordedSamples = append(recordedSamples, samples) - totalSamples += len(samples) - } - if len(stones) > 0 { - recordedDeletes = append(recordedDeletes, stones) - } - } - - require.NoError(t, w.Close()) - } -} - -func TestWALRestoreCorrupted_invalidSegment(t *testing.T) { - dir := t.TempDir() - - wal, err := OpenSegmentWAL(dir, nil, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(wal) - - _, err = wal.createSegmentFile(filepath.Join(dir, "000000")) - require.NoError(t, err) - f, err := wal.createSegmentFile(filepath.Join(dir, "000001")) - require.NoError(t, err) - f2, err := wal.createSegmentFile(filepath.Join(dir, "000002")) - require.NoError(t, err) - require.NoError(t, f2.Close()) - - // Make header of second segment invalid. - _, err = f.WriteAt([]byte{1, 2, 3, 4}, 0) - require.NoError(t, err) - require.NoError(t, f.Close()) - - require.NoError(t, wal.Close()) - - wal, err = OpenSegmentWAL(dir, log.NewLogfmtLogger(os.Stderr), 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(wal) - - files, err := os.ReadDir(dir) - require.NoError(t, err) - fns := []string{} - for _, f := range files { - fns = append(fns, f.Name()) - } - require.Equal(t, []string{"000000"}, fns) -} - -// Test reading from a WAL that has been corrupted through various means. -func TestWALRestoreCorrupted(t *testing.T) { - cases := []struct { - name string - f func(*testing.T, *SegmentWAL) - }{ - { - name: "truncate_checksum", - f: func(t *testing.T, w *SegmentWAL) { - f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0o666) - require.NoError(t, err) - defer f.Close() - - off, err := f.Seek(0, io.SeekEnd) - require.NoError(t, err) - - require.NoError(t, f.Truncate(off-1)) - }, - }, - { - name: "truncate_body", - f: func(t *testing.T, w *SegmentWAL) { - f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0o666) - require.NoError(t, err) - defer f.Close() - - off, err := f.Seek(0, io.SeekEnd) - require.NoError(t, err) - - require.NoError(t, f.Truncate(off-8)) - }, - }, - { - name: "body_content", - f: func(t *testing.T, w *SegmentWAL) { - f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0o666) - require.NoError(t, err) - defer f.Close() - - off, err := f.Seek(0, io.SeekEnd) - require.NoError(t, err) - - // Write junk before checksum starts. - _, err = f.WriteAt([]byte{1, 2, 3, 4}, off-8) - require.NoError(t, err) - }, - }, - { - name: "checksum", - f: func(t *testing.T, w *SegmentWAL) { - f, err := os.OpenFile(w.files[0].Name(), os.O_WRONLY, 0o666) - require.NoError(t, err) - defer f.Close() - - off, err := f.Seek(0, io.SeekEnd) - require.NoError(t, err) - - // Write junk into checksum - _, err = f.WriteAt([]byte{1, 2, 3, 4}, off-4) - require.NoError(t, err) - }, - }, - } - for _, c := range cases { - t.Run(c.name, func(t *testing.T) { - // Generate testing data. It does not make semantic sense but - // for the purpose of this test. - dir := t.TempDir() - - w, err := OpenSegmentWAL(dir, nil, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(w) - - require.NoError(t, w.LogSamples([]record.RefSample{{T: 1, V: 2}})) - require.NoError(t, w.LogSamples([]record.RefSample{{T: 2, V: 3}})) - - require.NoError(t, w.cut()) - - // Sleep 2 seconds to avoid error where cut and test "cases" function may write or - // truncate the file out of orders as "cases" are not synchronized with cut. - // Hopefully cut will complete by 2 seconds. - time.Sleep(2 * time.Second) - - require.NoError(t, w.LogSamples([]record.RefSample{{T: 3, V: 4}})) - require.NoError(t, w.LogSamples([]record.RefSample{{T: 5, V: 6}})) - - require.NoError(t, w.Close()) - - // cut() truncates and fsyncs the first segment async. If it happens after - // the corruption we apply below, the corruption will be overwritten again. - // Fire and forget a sync to avoid flakiness. - w.files[0].Sync() - // Corrupt the second entry in the first file. - // After re-opening we must be able to read the first entry - // and the rest, including the second file, must be truncated for clean further - // writes. - c.f(t, w) - - logger := log.NewLogfmtLogger(os.Stderr) - - w2, err := OpenSegmentWAL(dir, logger, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(w2) - - r := w2.Reader() - - serf := func(l []record.RefSeries) { - require.Empty(t, l) - } - - // Weird hack to check order of reads. - i := 0 - samplef := func(s []record.RefSample) { - if i == 0 { - require.Equal(t, []record.RefSample{{T: 1, V: 2}}, s) - i++ - } else { - require.Equal(t, []record.RefSample{{T: 99, V: 100}}, s) - } - } - - require.NoError(t, r.Read(serf, samplef, nil)) - - require.NoError(t, w2.LogSamples([]record.RefSample{{T: 99, V: 100}})) - require.NoError(t, w2.Close()) - - // We should see the first valid entry and the new one, everything after - // is truncated. - w3, err := OpenSegmentWAL(dir, logger, 0, nil) - require.NoError(t, err) - defer func(wal *SegmentWAL) { require.NoError(t, wal.Close()) }(w3) - - r = w3.Reader() - - i = 0 - require.NoError(t, r.Read(serf, samplef, nil)) - }) - } -} - -func TestMigrateWAL_Empty(t *testing.T) { - // The migration procedure must properly deal with a zero-length segment, - // which is valid in the new format. - dir := t.TempDir() - - wdir := path.Join(dir, "wal") - - // Initialize empty WAL. - w, err := wlog.New(nil, nil, wdir, wlog.CompressionNone) - require.NoError(t, err) - require.NoError(t, w.Close()) - - require.NoError(t, MigrateWAL(nil, wdir)) -} - -func TestMigrateWAL_Fuzz(t *testing.T) { - dir := t.TempDir() - - wdir := path.Join(dir, "wal") - - // Should pass if no WAL exists yet. - require.NoError(t, MigrateWAL(nil, wdir)) - - oldWAL, err := OpenSegmentWAL(wdir, nil, time.Minute, nil) - require.NoError(t, err) - - // Write some data. - require.NoError(t, oldWAL.LogSeries([]record.RefSeries{ - {Ref: 100, Labels: labels.FromStrings("abc", "def", "123", "456")}, - {Ref: 1, Labels: labels.FromStrings("abc", "def2", "1234", "4567")}, - })) - require.NoError(t, oldWAL.LogSamples([]record.RefSample{ - {Ref: 1, T: 100, V: 200}, - {Ref: 2, T: 300, V: 400}, - })) - require.NoError(t, oldWAL.LogSeries([]record.RefSeries{ - {Ref: 200, Labels: labels.FromStrings("xyz", "def", "foo", "bar")}, - })) - require.NoError(t, oldWAL.LogSamples([]record.RefSample{ - {Ref: 3, T: 100, V: 200}, - {Ref: 4, T: 300, V: 400}, - })) - require.NoError(t, oldWAL.LogDeletes([]tombstones.Stone{ - {Ref: 1, Intervals: []tombstones.Interval{{Mint: 100, Maxt: 200}}}, - })) - - require.NoError(t, oldWAL.Close()) - - // Perform migration. - require.NoError(t, MigrateWAL(nil, wdir)) - - w, err := wlog.New(nil, nil, wdir, wlog.CompressionNone) - require.NoError(t, err) - - // We can properly write some new data after migration. - var enc record.Encoder - require.NoError(t, w.Log(enc.Samples([]record.RefSample{ - {Ref: 500, T: 1, V: 1}, - }, nil))) - - require.NoError(t, w.Close()) - - // Read back all data. - sr, err := wlog.NewSegmentsReader(wdir) - require.NoError(t, err) - - r := wlog.NewReader(sr) - var res []interface{} - dec := record.NewDecoder(labels.NewSymbolTable()) - - for r.Next() { - rec := r.Record() - - switch dec.Type(rec) { - case record.Series: - s, err := dec.Series(rec, nil) - require.NoError(t, err) - res = append(res, s) - case record.Samples: - s, err := dec.Samples(rec, nil) - require.NoError(t, err) - res = append(res, s) - case record.Tombstones: - s, err := dec.Tombstones(rec, nil) - require.NoError(t, err) - res = append(res, s) - default: - require.Fail(t, "unknown record type %d", dec.Type(rec)) - } - } - require.NoError(t, r.Err()) - - testutil.RequireEqual(t, []interface{}{ - []record.RefSeries{ - {Ref: 100, Labels: labels.FromStrings("abc", "def", "123", "456")}, - {Ref: 1, Labels: labels.FromStrings("abc", "def2", "1234", "4567")}, - }, - []record.RefSample{{Ref: 1, T: 100, V: 200}, {Ref: 2, T: 300, V: 400}}, - []record.RefSeries{ - {Ref: 200, Labels: labels.FromStrings("xyz", "def", "foo", "bar")}, - }, - []record.RefSample{{Ref: 3, T: 100, V: 200}, {Ref: 4, T: 300, V: 400}}, - []tombstones.Stone{{Ref: 1, Intervals: []tombstones.Interval{{Mint: 100, Maxt: 200}}}}, - []record.RefSample{{Ref: 500, T: 1, V: 1}}, - }, res) - - // Migrating an already migrated WAL shouldn't do anything. - require.NoError(t, MigrateWAL(nil, wdir)) -} diff --git a/tsdb/wlog/checkpoint.go b/tsdb/wlog/checkpoint.go index 4ad1bb2365..a16cd5fc74 100644 --- a/tsdb/wlog/checkpoint.go +++ b/tsdb/wlog/checkpoint.go @@ -149,22 +149,23 @@ func Checkpoint(logger log.Logger, w *WL, from, to int, keep func(id chunks.Head r := NewReader(sgmReader) var ( - series []record.RefSeries - samples []record.RefSample - histogramSamples []record.RefHistogramSample - tstones []tombstones.Stone - exemplars []record.RefExemplar - metadata []record.RefMetadata - st = labels.NewSymbolTable() // Needed for decoding; labels do not outlive this function. - dec = record.NewDecoder(st) - enc record.Encoder - buf []byte - recs [][]byte + series []record.RefSeries + samples []record.RefSample + histogramSamples []record.RefHistogramSample + floatHistogramSamples []record.RefFloatHistogramSample + tstones []tombstones.Stone + exemplars []record.RefExemplar + metadata []record.RefMetadata + st = labels.NewSymbolTable() // Needed for decoding; labels do not outlive this function. + dec = record.NewDecoder(st) + enc record.Encoder + buf []byte + recs [][]byte latestMetadataMap = make(map[chunks.HeadSeriesRef]record.RefMetadata) ) for r.Next() { - series, samples, histogramSamples, tstones, exemplars, metadata = series[:0], samples[:0], histogramSamples[:0], tstones[:0], exemplars[:0], metadata[:0] + series, samples, histogramSamples, floatHistogramSamples, tstones, exemplars, metadata = series[:0], samples[:0], histogramSamples[:0], floatHistogramSamples[:0], tstones[:0], exemplars[:0], metadata[:0] // We don't reset the buffer since we batch up multiple records // before writing them to the checkpoint. @@ -224,8 +225,26 @@ func Checkpoint(logger log.Logger, w *WL, from, to int, keep func(id chunks.Head if len(repl) > 0 { buf = enc.HistogramSamples(repl, buf) } - stats.TotalSamples += len(samples) - stats.DroppedSamples += len(samples) - len(repl) + stats.TotalSamples += len(histogramSamples) + stats.DroppedSamples += len(histogramSamples) - len(repl) + + case record.FloatHistogramSamples: + floatHistogramSamples, err = dec.FloatHistogramSamples(rec, floatHistogramSamples) + if err != nil { + return nil, fmt.Errorf("decode float histogram samples: %w", err) + } + // Drop irrelevant floatHistogramSamples in place. + repl := floatHistogramSamples[:0] + for _, fh := range floatHistogramSamples { + if fh.T >= mint { + repl = append(repl, fh) + } + } + if len(repl) > 0 { + buf = enc.FloatHistogramSamples(repl, buf) + } + stats.TotalSamples += len(floatHistogramSamples) + stats.DroppedSamples += len(floatHistogramSamples) - len(repl) case record.Tombstones: tstones, err = dec.Tombstones(rec, tstones) diff --git a/tsdb/wlog/checkpoint_test.go b/tsdb/wlog/checkpoint_test.go index 142a5a9d49..279f7c4356 100644 --- a/tsdb/wlog/checkpoint_test.go +++ b/tsdb/wlog/checkpoint_test.go @@ -125,6 +125,20 @@ func TestCheckpoint(t *testing.T) { PositiveBuckets: []int64{int64(i + 1), 1, -1, 0}, } } + makeFloatHistogram := func(i int) *histogram.FloatHistogram { + return &histogram.FloatHistogram{ + Count: 5 + float64(i*4), + ZeroCount: 2 + float64(i), + ZeroThreshold: 0.001, + Sum: 18.4 * float64(i+1), + Schema: 1, + PositiveSpans: []histogram.Span{ + {Offset: 0, Length: 2}, + {Offset: 1, Length: 2}, + }, + PositiveBuckets: []float64{float64(i + 1), 1, -1, 0}, + } + } for _, compress := range []CompressionType{CompressionNone, CompressionSnappy, CompressionZstd} { t.Run(fmt.Sprintf("compress=%s", compress), func(t *testing.T) { @@ -154,7 +168,7 @@ func TestCheckpoint(t *testing.T) { w, err = NewSize(nil, nil, dir, 64*1024, compress) require.NoError(t, err) - samplesInWAL, histogramsInWAL := 0, 0 + samplesInWAL, histogramsInWAL, floatHistogramsInWAL := 0, 0, 0 var last int64 for i := 0; ; i++ { _, n, err := Segments(w.Dir()) @@ -200,6 +214,15 @@ func TestCheckpoint(t *testing.T) { }, nil) require.NoError(t, w.Log(b)) histogramsInWAL += 4 + fh := makeFloatHistogram(i) + b = enc.FloatHistogramSamples([]record.RefFloatHistogramSample{ + {Ref: 0, T: last, FH: fh}, + {Ref: 1, T: last + 10000, FH: fh}, + {Ref: 2, T: last + 20000, FH: fh}, + {Ref: 3, T: last + 30000, FH: fh}, + }, nil) + require.NoError(t, w.Log(b)) + floatHistogramsInWAL += 4 b = enc.Exemplars([]record.RefExemplar{ {Ref: 1, T: last, V: float64(i), Labels: labels.FromStrings("trace_id", fmt.Sprintf("trace-%d", i))}, @@ -220,12 +243,14 @@ func TestCheckpoint(t *testing.T) { } require.NoError(t, w.Close()) - _, err = Checkpoint(log.NewNopLogger(), w, 100, 106, func(x chunks.HeadSeriesRef) bool { + stats, err := Checkpoint(log.NewNopLogger(), w, 100, 106, func(x chunks.HeadSeriesRef) bool { return x%2 == 0 }, last/2) require.NoError(t, err) require.NoError(t, w.Truncate(107)) require.NoError(t, DeleteCheckpoints(w.Dir(), 106)) + require.Equal(t, histogramsInWAL+floatHistogramsInWAL+samplesInWAL, stats.TotalSamples) + require.Greater(t, stats.DroppedSamples, 0) // Only the new checkpoint should be left. files, err := os.ReadDir(dir) @@ -242,7 +267,7 @@ func TestCheckpoint(t *testing.T) { var metadata []record.RefMetadata r := NewReader(sr) - samplesInCheckpoint, histogramsInCheckpoint := 0, 0 + samplesInCheckpoint, histogramsInCheckpoint, floatHistogramsInCheckpoint := 0, 0, 0 for r.Next() { rec := r.Record() @@ -264,6 +289,13 @@ func TestCheckpoint(t *testing.T) { require.GreaterOrEqual(t, h.T, last/2, "histogram with wrong timestamp") } histogramsInCheckpoint += len(histograms) + case record.FloatHistogramSamples: + floatHistograms, err := dec.FloatHistogramSamples(rec, nil) + require.NoError(t, err) + for _, h := range floatHistograms { + require.GreaterOrEqual(t, h.T, last/2, "float histogram with wrong timestamp") + } + floatHistogramsInCheckpoint += len(floatHistograms) case record.Exemplars: exemplars, err := dec.Exemplars(rec, nil) require.NoError(t, err) @@ -281,6 +313,8 @@ func TestCheckpoint(t *testing.T) { require.Less(t, float64(samplesInCheckpoint)/float64(samplesInWAL), 0.8) require.Greater(t, float64(histogramsInCheckpoint)/float64(histogramsInWAL), 0.5) require.Less(t, float64(histogramsInCheckpoint)/float64(histogramsInWAL), 0.8) + require.Greater(t, float64(floatHistogramsInCheckpoint)/float64(floatHistogramsInWAL), 0.5) + require.Less(t, float64(floatHistogramsInCheckpoint)/float64(floatHistogramsInWAL), 0.8) expectedRefSeries := []record.RefSeries{ {Ref: 0, Labels: labels.FromStrings("a", "b", "c", "0")}, diff --git a/tsdb/wlog/live_reader.go b/tsdb/wlog/live_reader.go index 905bbf00d6..6eaef5f396 100644 --- a/tsdb/wlog/live_reader.go +++ b/tsdb/wlog/live_reader.go @@ -327,10 +327,3 @@ func (r *LiveReader) readRecord() ([]byte, int, error) { return rec, length + recordHeaderSize, nil } - -func min(i, j int) int { - if i < j { - return i - } - return j -} diff --git a/util/httputil/compression_test.go b/util/httputil/compression_test.go index 2db6810bd7..e166c7de79 100644 --- a/util/httputil/compression_test.go +++ b/util/httputil/compression_test.go @@ -85,7 +85,7 @@ func TestCompressionHandler_Gzip(t *testing.T) { }, } - req, _ := http.NewRequest("GET", server.URL+"/foo_endpoint", nil) + req, _ := http.NewRequest(http.MethodGet, server.URL+"/foo_endpoint", nil) req.Header.Set(acceptEncodingHeader, gzipEncoding) resp, err := client.Do(req) @@ -120,7 +120,7 @@ func TestCompressionHandler_Deflate(t *testing.T) { }, } - req, _ := http.NewRequest("GET", server.URL+"/foo_endpoint", nil) + req, _ := http.NewRequest(http.MethodGet, server.URL+"/foo_endpoint", nil) req.Header.Set(acceptEncodingHeader, deflateEncoding) resp, err := client.Do(req) diff --git a/util/httputil/cors_test.go b/util/httputil/cors_test.go index cfa2400405..657443ece0 100644 --- a/util/httputil/cors_test.go +++ b/util/httputil/cors_test.go @@ -41,7 +41,7 @@ func TestCORSHandler(t *testing.T) { dummyOrigin := "https://foo.com" // OPTIONS with legit origin - req, err := http.NewRequest("OPTIONS", server.URL+"/any_path", nil) + req, err := http.NewRequest(http.MethodOptions, server.URL+"/any_path", nil) require.NoError(t, err, "could not create request") req.Header.Set("Origin", dummyOrigin) @@ -53,7 +53,7 @@ func TestCORSHandler(t *testing.T) { require.Equal(t, dummyOrigin, AccessControlAllowOrigin, "expected Access-Control-Allow-Origin header") // OPTIONS with bad origin - req, err = http.NewRequest("OPTIONS", server.URL+"/any_path", nil) + req, err = http.NewRequest(http.MethodOptions, server.URL+"/any_path", nil) require.NoError(t, err, "could not create request") req.Header.Set("Origin", "https://not-foo.com") diff --git a/util/teststorage/storage.go b/util/teststorage/storage.go index 5d95437e99..7d1f9dda24 100644 --- a/util/teststorage/storage.go +++ b/util/teststorage/storage.go @@ -14,6 +14,7 @@ package teststorage import ( + "fmt" "os" "time" @@ -30,8 +31,18 @@ import ( // New returns a new TestStorage for testing purposes // that removes all associated files on closing. func New(t testutil.T) *TestStorage { + stor, err := NewWithError() + require.NoError(t, err) + return stor +} + +// NewWithError returns a new TestStorage for user facing tests, which reports +// errors directly. +func NewWithError() (*TestStorage, error) { dir, err := os.MkdirTemp("", "test_storage") - require.NoError(t, err, "unexpected error while opening test directory") + if err != nil { + return nil, fmt.Errorf("opening test directory: %w", err) + } // Tests just load data for a series sequentially. Thus we // need a long appendable window. @@ -41,13 +52,17 @@ func New(t testutil.T) *TestStorage { opts.RetentionDuration = 0 opts.EnableNativeHistograms = true db, err := tsdb.Open(dir, nil, nil, opts, tsdb.NewDBStats()) - require.NoError(t, err, "unexpected error while opening test storage") + if err != nil { + return nil, fmt.Errorf("opening test storage: %w", err) + } reg := prometheus.NewRegistry() eMetrics := tsdb.NewExemplarMetrics(reg) es, err := tsdb.NewCircularExemplarStorage(10, eMetrics) - require.NoError(t, err, "unexpected error while opening test exemplar storage") - return &TestStorage{DB: db, exemplarStorage: es, dir: dir} + if err != nil { + return nil, fmt.Errorf("opening test exemplar storage: %w", err) + } + return &TestStorage{DB: db, exemplarStorage: es, dir: dir}, nil } type TestStorage struct { diff --git a/web/api/v1/api.go b/web/api/v1/api.go index 48938fce12..dc22365073 100644 --- a/web/api/v1/api.go +++ b/web/api/v1/api.go @@ -177,13 +177,6 @@ type TSDBAdminStats interface { WALReplayStatus() (tsdb.WALReplayStatus, error) } -// QueryEngine defines the interface for the *promql.Engine, so it can be replaced, wrapped or mocked. -type QueryEngine interface { - SetQueryLogger(l promql.QueryLogger) - NewInstantQuery(ctx context.Context, q storage.Queryable, opts promql.QueryOpts, qs string, ts time.Time) (promql.Query, error) - NewRangeQuery(ctx context.Context, q storage.Queryable, opts promql.QueryOpts, qs string, start, end time.Time, interval time.Duration) (promql.Query, error) -} - type QueryOpts interface { EnablePerStepStats() bool LookbackDelta() time.Duration @@ -193,7 +186,7 @@ type QueryOpts interface { // them using the provided storage and query engine. type API struct { Queryable storage.SampleAndChunkQueryable - QueryEngine QueryEngine + QueryEngine promql.QueryEngine ExemplarQueryable storage.ExemplarQueryable scrapePoolsRetriever func(context.Context) ScrapePoolsRetriever @@ -226,7 +219,7 @@ type API struct { // NewAPI returns an initialized API type. func NewAPI( - qe QueryEngine, + qe promql.QueryEngine, q storage.SampleAndChunkQueryable, ap storage.Appendable, eq storage.ExemplarQueryable, @@ -889,6 +882,9 @@ func (api *API) series(r *http.Request) (result apiFuncResult) { warnings := set.Warnings() for set.Next() { + if err := ctx.Err(); err != nil { + return apiFuncResult{nil, returnAPIError(err), warnings, closer} + } metrics = append(metrics, set.At().Labels()) if len(metrics) >= limit { diff --git a/web/api/v1/api_test.go b/web/api/v1/api_test.go index 6ccfdb5d7f..c383993815 100644 --- a/web/api/v1/api_test.go +++ b/web/api/v1/api_test.go @@ -3354,7 +3354,7 @@ func TestParseTimeParam(t *testing.T) { } for _, test := range tests { - req, err := http.NewRequest("GET", "localhost:42/foo?"+test.paramName+"="+test.paramValue, nil) + req, err := http.NewRequest(http.MethodGet, "localhost:42/foo?"+test.paramName+"="+test.paramValue, nil) require.NoError(t, err) result := test.result @@ -3491,7 +3491,7 @@ func TestOptionsMethod(t *testing.T) { s := httptest.NewServer(r) defer s.Close() - req, err := http.NewRequest("OPTIONS", s.URL+"/any_path", nil) + req, err := http.NewRequest(http.MethodOptions, s.URL+"/any_path", nil) require.NoError(t, err, "Error creating OPTIONS request") client := &http.Client{} resp, err := client.Do(req) @@ -3568,6 +3568,9 @@ func TestReturnAPIError(t *testing.T) { }, { err: errors.New("exec error"), expected: errorExec, + }, { + err: context.Canceled, + expected: errorCanceled, }, } @@ -3881,8 +3884,6 @@ type fakeEngine struct { query fakeQuery } -func (e *fakeEngine) SetQueryLogger(promql.QueryLogger) {} - func (e *fakeEngine) NewInstantQuery(ctx context.Context, q storage.Queryable, opts promql.QueryOpts, qs string, ts time.Time) (promql.Query, error) { return &e.query, nil } diff --git a/web/api/v1/errors_test.go b/web/api/v1/errors_test.go index b6ec7d4e1f..e76a1a3d35 100644 --- a/web/api/v1/errors_test.go +++ b/web/api/v1/errors_test.go @@ -89,7 +89,7 @@ func TestApiStatusCodes(t *testing.T) { r := createPrometheusAPI(q) rec := httptest.NewRecorder() - req := httptest.NewRequest("GET", "/api/v1/query?query=up", nil) + req := httptest.NewRequest(http.MethodGet, "/api/v1/query?query=up", nil) r.ServeHTTP(rec, req) diff --git a/web/federate_test.go b/web/federate_test.go index b8749dfa32..f201210ec0 100644 --- a/web/federate_test.go +++ b/web/federate_test.go @@ -48,29 +48,29 @@ var scenarios = map[string]struct { }{ "empty": { params: "", - code: 200, + code: http.StatusOK, body: ``, }, "match nothing": { params: "match[]=does_not_match_anything", - code: 200, + code: http.StatusOK, body: ``, }, "invalid params from the beginning": { params: "match[]=-not-a-valid-metric-name", - code: 400, + code: http.StatusBadRequest, body: `1:1: parse error: unexpected `, }, "invalid params somewhere in the middle": { params: "match[]=not-a-valid-metric-name", - code: 400, + code: http.StatusBadRequest, body: `1:4: parse error: unexpected `, }, "test_metric1": { params: "match[]=test_metric1", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -78,33 +78,33 @@ test_metric1{foo="boo",instance="i"} 1 6000000 }, "test_metric2": { params: "match[]=test_metric2", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric2 untyped test_metric2{foo="boo",instance="i"} 1 6000000 `, }, "test_metric_without_labels": { params: "match[]=test_metric_without_labels", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric_without_labels untyped test_metric_without_labels{instance=""} 1001 6000000 `, }, "test_stale_metric": { params: "match[]=test_metric_stale", - code: 200, + code: http.StatusOK, body: ``, }, "test_old_metric": { params: "match[]=test_metric_old", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric_old untyped test_metric_old{instance=""} 981 5880000 `, }, "{foo='boo'}": { params: "match[]={foo='boo'}", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="boo",instance="i"} 1 6000000 # TYPE test_metric2 untyped @@ -113,7 +113,7 @@ test_metric2{foo="boo",instance="i"} 1 6000000 }, "two matchers": { params: "match[]=test_metric1&match[]=test_metric2", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -123,7 +123,7 @@ test_metric2{foo="boo",instance="i"} 1 6000000 }, "two matchers with overlap": { params: "match[]={__name__=~'test_metric1'}&match[]={foo='bar'}", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -131,7 +131,7 @@ test_metric1{foo="boo",instance="i"} 1 6000000 }, "everything": { params: "match[]={__name__=~'.%2b'}", // '%2b' is an URL-encoded '+'. - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -145,7 +145,7 @@ test_metric_without_labels{instance=""} 1001 6000000 }, "empty label value matches everything that doesn't have that label": { params: "match[]={foo='',__name__=~'.%2b'}", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric_old untyped test_metric_old{instance=""} 981 5880000 # TYPE test_metric_without_labels untyped @@ -154,7 +154,7 @@ test_metric_without_labels{instance=""} 1001 6000000 }, "empty label value for a label that doesn't exist at all, matches everything": { params: "match[]={bar='',__name__=~'.%2b'}", - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -169,7 +169,7 @@ test_metric_without_labels{instance=""} 1001 6000000 "external labels are added if not already present": { params: "match[]={__name__=~'.%2b'}", // '%2b' is an URL-encoded '+'. externalLabels: labels.FromStrings("foo", "baz", "zone", "ie"), - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i",zone="ie"} 10000 6000000 test_metric1{foo="boo",instance="i",zone="ie"} 1 6000000 @@ -186,7 +186,7 @@ test_metric_without_labels{foo="baz",instance="",zone="ie"} 1001 6000000 // know what it does anyway. params: "match[]={__name__=~'.%2b'}", // '%2b' is an URL-encoded '+'. externalLabels: labels.FromStrings("instance", "baz"), - code: 200, + code: http.StatusOK, body: `# TYPE test_metric1 untyped test_metric1{foo="bar",instance="i"} 10000 6000000 test_metric1{foo="boo",instance="i"} 1 6000000 @@ -224,7 +224,7 @@ func TestFederation(t *testing.T) { for name, scenario := range scenarios { t.Run(name, func(t *testing.T) { h.config.GlobalConfig.ExternalLabels = scenario.externalLabels - req := httptest.NewRequest("GET", "http://example.org/federate?"+scenario.params, nil) + req := httptest.NewRequest(http.MethodGet, "http://example.org/federate?"+scenario.params, nil) res := httptest.NewRecorder() h.federation(res, req) @@ -265,7 +265,7 @@ func TestFederation_NotReady(t *testing.T) { }, } - req := httptest.NewRequest("GET", "http://example.org/federate?"+scenario.params, nil) + req := httptest.NewRequest(http.MethodGet, "http://example.org/federate?"+scenario.params, nil) res := httptest.NewRecorder() h.federation(res, req) @@ -381,7 +381,7 @@ func TestFederationWithNativeHistograms(t *testing.T) { }, } - req := httptest.NewRequest("GET", "http://example.org/federate?match[]=test_metric", nil) + req := httptest.NewRequest(http.MethodGet, "http://example.org/federate?match[]=test_metric", nil) req.Header.Add("Accept", `application/vnd.google.protobuf;proto=io.prometheus.client.MetricFamily;encoding=delimited,application/openmetrics-text;version=1.0.0;q=0.8,application/openmetrics-text;version=0.0.1;q=0.75,text/plain;version=0.0.4;q=0.5,*/*;q=0.1`) res := httptest.NewRecorder() @@ -390,7 +390,6 @@ func TestFederationWithNativeHistograms(t *testing.T) { require.Equal(t, http.StatusOK, res.Code) body, err := io.ReadAll(res.Body) require.NoError(t, err) - p := textparse.NewProtobufParser(body, false, labels.NewSymbolTable()) var actVec promql.Vector metricFamilies := 0 diff --git a/web/web_test.go b/web/web_test.go index 62bdb2ae31..e1fa66fa8b 100644 --- a/web/web_test.go +++ b/web/web_test.go @@ -311,7 +311,7 @@ func TestDebugHandler(t *testing.T) { w := httptest.NewRecorder() - req, err := http.NewRequest("GET", tc.url, nil) + req, err := http.NewRequest(http.MethodGet, tc.url, nil) require.NoError(t, err) @@ -335,7 +335,7 @@ func TestHTTPMetrics(t *testing.T) { t.Helper() w := httptest.NewRecorder() - req, err := http.NewRequest("GET", "/-/ready", nil) + req, err := http.NewRequest(http.MethodGet, "/-/ready", nil) require.NoError(t, err) handler.router.ServeHTTP(w, req)