Implements running plugins in containers to give them some degree
of isolation from the main Vault process and other plugins. It only
supports running on Linux initially, where it is easiest to manage unix
socket communication across the container boundary.
Additionally
* Adds -env arg to vault plugin register.
* Don't return env from 'vault plugin info'
Historically it's been omitted, and it could conceivably have secret information in
it, so if we want to return it in the response, it should probably only be via explicit
opt-in. Skipping for now though as it's not the main purpose of the commit.
The Host part of the URL doesn't actually get used when we initiate connections to UNIX domain sockets. As of https://github.com/golang/go/issues/60374 (in the latest Go releases at the time of this writing), we must set it to something that looks like a hostname or requests will fail.
* Adding explicit MPL license for sub-package.
This directory and its subdirectories (packages) contain files licensed with the MPLv2 `LICENSE` file in this directory and are intentionally licensed separately from the BSL `LICENSE` file at the root of this repository.
* Adding explicit MPL license for sub-package.
This directory and its subdirectories (packages) contain files licensed with the MPLv2 `LICENSE` file in this directory and are intentionally licensed separately from the BSL `LICENSE` file at the root of this repository.
* Updating the license from MPL to Business Source License.
Going forward, this project will be licensed under the Business Source License v1.1. Please see our blog post for more details at https://hashi.co/bsl-blog, FAQ at www.hashicorp.com/licensing-faq, and details of the license at www.hashicorp.com/bsl.
* add missing license headers
* Update copyright file headers to BUS-1.1
* Fix test that expected exact offset on hcl file
---------
Co-authored-by: hashicorp-copywrite[bot] <110428419+hashicorp-copywrite[bot]@users.noreply.github.com>
Co-authored-by: Sarah Thompson <sthompson@hashicorp.com>
Co-authored-by: Brian Kassouf <bkassouf@hashicorp.com>
* Fix regexes for `sys/raw/` and `sys/leases/lookup/` to match prevailing conventions
There are several endpoints in Vault which take an arbitrary path as the
last parameter. Many of these are defined in terms of the
`framework.MatchAllRegex` helper. Some were not, and were defined using
custom regexes which gave rise to multiple OpenAPI endpoints - one with
the path parameter, and one without.
We need to fix these definitions, because they give rise to a very
unnatural result when used to generate a client API - for example, you
end up with `LeasesLookUp()` which is only capable of being used to list
the very top level of the hierarchical collection of leases, and
`LeasesLookUpWithPrefix(prefix)` which must be used for all deeper
levels.
This PR changes the regexes used for `sys/raw/` and `sys/leases/lookup/`
to be consistent with the approach used for other well-known similar
endpoints, such as `cubbyhole/`, `kv-v1/` and `kv-v2/metadata/`.
This PR does have a very small compatibility issue, which I think is
tolerable: prior to this change, `sys/raw` with no trailing slash was
considered a valid endpoint, and now it will no longer be.
One way to observe this is to try `vault path-help sys/raw` - before
this change, it would work, after, it will not. You would have to
instead use `vault path-help sys/raw/foobar` to see the help.
I also considered whether losing the ability to read/write/delete
`sys/raw` would be an issue. In each case, the precise HTTP result code
will change, but each of these were meaningless operations that make no
sense - you cannot read/write/delete a "file" at the "root directory" of
the underlying Vault storage.
In fact, during testing, I discovered that currently, `vault write
sys/raw x=y` when using Raft storage, will permanently break the Vault
instance - it causes a panic within the Raft FSM, which re-occurs
immediately on restarting the server! This PR also closes off that
footgun / DoS vector.
None of these issues apply to `sys/leases/lookup/`, as the existing
regex in that case was already not matching the path without the
trailing slash.
* changelog
* Realign hardcoded sudo paths with updated OpenAPI spec
* Fix sudo paths missing from OpenAPI and docs
Various sudo (a.k.a. root-protected) paths are implemented in
non-standard ways, and as a result:
* are not declared as x-vault-sudo in the OpenAPI spec
* and as a result of that, are not included in the hardcoded patterns
powering the Vault CLI `-output-policy` flag
* and in some cases are missing from the table of all sudo paths in the
docs too
Fix these problems by:
* Adding `seal` and `step-down` to the list of root paths for the system
backend. They don't need to be there for enforcement, as those two
special endpoints bypass the standard request handling code, but they
do need to be there for the OpenAPI generator to be able to know they
require sudo.
The way in which those two endpoints do things differently can be
observed in the code search results for `RootPrivsRequired`:
https://github.com/search?q=repo%3Ahashicorp%2Fvault%20RootPrivsRequired&type=code
* Fix the implementation of `auth/token/revoke-orphan` to implement
endpoint sudo requirements in the standard way. Currently, it has an
**incorrect** path declared in the special paths metadata, and then
compensates with custom code throwing an error within the request
handler function itself.
* changelog
* As discussed in PR, delete test which is just testing equality of a constant
* Restore sudo check as requested, and add comment
* Update vault/token_store.go
Co-authored-by: Anton Averchenkov <84287187+averche@users.noreply.github.com>
---------
Co-authored-by: Anton Averchenkov <84287187+averche@users.noreply.github.com>
This is a code cleanup and addition of an explanatory comment. For some
reason, the code related to the CLI guessing whether a path requires
sudo, has been interleaved into plugin_helpers.go, which was previously
purely code used on the server side in the implementation of Vault
plugins.
This remedies that by dividing the sudo paths code to a separate file,
and adds a comment to plugin_helpers.go providing future readers with
information about the overall theme of the file.
No code has been changed - only moved and documented.
* OpenAPI: Separate ListOperation from ReadOperation
Historically, since Vault's ReadOperation and ListOperation both map to
the HTTP GET method, their representation in the generated OpenAPI has
been a bit confusing.
This was partially mitigated some time ago, by making the `list` query
parameter express whether it was required or optional - but only in
a way useful to human readers - the human had to know, for example, that
the schema of the response body would change depending on whether `list`
was selected.
Now that there is an effort underway to automatically generate API
clients from the OpenAPI spec, we have a need to fix this more
comprehensively. Fortunately, we do have a means to do so - since Vault
has opinionated treatment of trailing slashes, linked to operations
being list or not, we can use an added trailing slash on the URL path to
separate list operations in the OpenAPI spec.
This PR implements that, and then fixes an operation ID which becomes
duplicated, with this change applied.
See also hashicorp/vault-client-go#174, a bug which will be fixed by
this work.
* Set further DisplayAttrs in auth/github plugin
To mask out more duplicate read/list functionality, now being separately
generated to OpenAPI client libraries as a result of this change.
* Apply requested changes to operation IDs
I'm not totally convinced its worth the extra lines of code, but
equally, I don't have strong feelings about it, so I'll just make the
change.
* Adjust logic to prevent any possibility of generating OpenAPI paths with doubled final slashes
Even in the edge case of improper use of regex patterns and operations.
* changelog
* Fix TestSudoPaths to pass again... which snowballed a bit...
Once I looked hard at it, I found it was missing several sudo paths,
which led to additional bug fixing elsewhere.
I might need to pull some parts of this change out into a separate PR
for ease of review...
* Fix other tests
* More test fixing
* Undo scope creep - back away from fixing sudo paths not shown as such in OpenAPI, at least within this PR
Just add TODO comments for now.
* Fix `vault path-help` for selected paths with bad regexps
See the comment being added in `sdk/framework/path.go` for the
explanation of why this change is needed.
* Grammar fix and add changelog
* Also fix hardcoded expectations in a new test
* Add a couple more testcases, and some comments.
* Tweak spelling in comment
Additional fixes:
- handle a failed type assert in api.Config.configureTLS()
Co-authored-by: Anton Averchenkov <84287187+averche@users.noreply.github.com>
* upgrade go-jose library to v3
Signed-off-by: Mark Sagi-Kazar <mark.sagikazar@gmail.com>
* chore: fix unnecessary import alias
Signed-off-by: Mark Sagi-Kazar <mark.sagikazar@gmail.com>
* upgrade go-jose library to v2 in vault
Signed-off-by: Mark Sagi-Kazar <mark.sagikazar@gmail.com>
---------
Signed-off-by: Mark Sagi-Kazar <mark.sagikazar@gmail.com>
* update error message and properly handle list requests
* since we do agressive sanitizes we need to optionally check trailing slash
* added changelog record
* remove redundant path formating
* Update changelog/13106.txt
Co-authored-by: Anton Averchenkov <84287187+averche@users.noreply.github.com>
* addressed comments from review
* also remove code that duplicates efforts in kv_list
* abstracted helper func for testing
* added test cases for the policy builder
* updated the changelog to the correct one
* removed calls that apear not to do anything given test case results
* fixed spacing issue in output string
* remove const representation of list url param
* addressed comments for pr
---------
Co-authored-by: lursu <leland.ursu@hashicorp.com>
Co-authored-by: Anton Averchenkov <84287187+averche@users.noreply.github.com>
* Revert "Don't execute the seal recovery tests on ENT. (#18841)"
This reverts commit 990d3bacc2.
* Revert "Add the ability to unseal using recovery keys via an explicit seal option. (#18683)"
This reverts commit 2ffe49aab0.
* wip
* wip
* Got it 'working', but not happy about cleanliness yet
* Switch to a dedicated defaultSeal with recovery keys
This is simpler than trying to hijack SealAccess as before. Instead, if the operator
has requested recovery unseal mode (via a flag in the seal stanza), we new up a shamir
seal with the recovery unseal key path instead of the auto seal. Then everything proceeds
as if you had a shamir seal to begin with.
* Handle recovery rekeying
* changelog
* Revert go.mod redirect
* revert multi-blob info
* Dumb nil unmarshal target
* More comments
* Update vault/seal.go
Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>
* Update changelog/18683.txt
Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>
* pr feedback
* Fix recovery rekey, which needs to fetch root keys and restore them under the new recovery split
* Better comment on recovery seal during adjustSealMigration
* Make it possible to migrate from an auto-seal in recovery mode to shamir
* Fix sealMigrated to account for a recovery seal
* comments
* Update changelog/18683.txt
Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>
* Address PR feedback
* Refactor duplicated migration code into helpers, using UnsealRecoveryKey/RecoveryKey where appropriate
* Don't shortcut the reast of seal migration
* get rid of redundant transit server cleanup
Co-authored-by: Nick Cabatoff <ncabatoff@hashicorp.com>
* Update golang.org/x/crypto version
go get -u golang.org/x/crypto && go mod tidy
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Update golang.org/x/crypto version in api
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Update golang.org/x/crypto version in sdk
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
Removing the timeout logic from raw-response functions and adding documentation comments. The following functions are affected:
- `ReadRaw`
- `ReadRawWithContext` (newly added)
- `ReadRawWithData`
- `ReadRawWithDataWithContext`
The previous logic of using `ctx, _ = c.c.withConfiguredTimeout(ctx)` could cause a potential [context leak](https://pkg.go.dev/context):
> Failing to call the CancelFunc leaks the child and its children until the parent is canceled or the timer fires. The go vet tool checks that CancelFuncs are used on all control-flow paths.
Cancelling the context would have caused more issues since the context would be cancelled before the request body is closed.
Resolves: #18658
* Revert "Add mount path into the default generated openapi.json spec (UI) (#17926)"
This reverts commit db8efac708.
* Revert "Remove `generic_mount_paths` field (#18558)"
This reverts commit 79c8f626c5.
* Stub out initial health check command
This command will be used to generate health check results for the PKI
engine.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Start common health check implementation
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Add common health check utilities
These utilities will collect helpers not specific to PKI health checks,
such as formatting longer durations more legibly.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Add PKI health check common utils
Many health checks will need issuer and/or CRL information in order to
execute. We've centrally located these helpers to avoid particular
health checks from needing to reimplement them each time.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Adding ca_validity_period health check
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Begin using health-checks in PKI command
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Allow parsing raw requests afterwards
This shifts the last of the logic difference between Read(...) and
ReadRaw(...) to a new helper, allowing ReadRaw(...) requests to be
parsed into the same response structure afterwards as Read(...); this
allows API callers to fetch the raw secret and inspect the raw response
object in case something went wrong (error code &c) -- and when the
request succeeds, they can still get the api.Secret out.
This will be used with the PKI health check functionality, making both
LIST and READ operations use ReadRaw, and optionally parsing the secret
afterwards.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Add crl_validity_period health check
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Add tests for PKI health check
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Fix bug in raw reading with contexts
When reading raw objects, don't manually call the context cancellation:
this causes timeouts and/or EOF errors when attempting to read or parse
the response body. See message in client.RawRequestWithContext(...) for
more information.
This was causing the test suite to randomly fail, due to the context
cancelling. The test suite's client usually had a default timeout,
whereas the CLI didn't, and thus didn't exhibit the same issue.
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Add changelog
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Fix typo in permissions message
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
* Move %v->%w for errs
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
Signed-off-by: Alexander Scheel <alex.scheel@hashicorp.com>
The current behaviour is to only add mount paths into the generated `opeanpi.json` spec if a `generic_mount_paths` flag is added to the request. This means that we would have to maintain two different `openapi.json` files, which is not ideal. The new solution in this PR is to add `{mount_path}` into every path with a default value specified:
```diff
-- "/auth/token/accessors/": {
++ "/auth/{mount_path}/accessors/": {
"parameters": [
{
"name": "mount_path",
"description": "....",
"in": "path",
"schema": {
"type": "string",
++ "default": "token"
}
}
],
```
Additionally, fixed the logic to generate the `operationId` (used to generate method names in the code generated from OpenAPI spec). It had a bug where the ID had `mountPath` in it. The new ID will look like this:
```diff
-- "operationId": "listAuthMountpathAccessors",
++ "operationId": "listTokenAccessors",
```
