When AAAA multi-target / dual stack support was
added via #2461 it broke ownership of domains across
different clusters with different ingress records types.
For example if 2 clusters manage the same zone,
1 cluster uses A records and the other uses CNAME
records, when each record type is treated as a separate
planning record, it will cause ownership to bounce back
and forth and records to be constantly created and
deleted.
This change updates the planner to keep track of multiple
current records for a domain. This allows for A and AAAA
records to exist for a domain while allowing record type
changes.
The planner will ignore desired records for a domain that
represent conflicting record types allowed by RFC 1034 3.6.2.
For example if the desired records for a domain contains
a CNAME record plus any other record type no changes for
that domain will be planned.
The planner now contains an owned record filter provided
by the registry. This allows the planner to accurately plan
create updates when there are record type changes between
the current and desired endpoints. Without this filter the
planner could add create changes for domains not owned
by the controller.
This change tries to remove part of the logic added in 50f196c0db.
The forceUpdate strategy relies on existing code of the planner to migrate TXT records to the new format,
being the main goal to avoid conflicts during apply.
Signed-off-by: Matias Charriere <matias@giantswarm.io>
Currently, planning instructs to create all records even
those which does not match any zone.
Later, those records will be checked towards the existing
records and filtered whether they match or not a hosted zone.
This causes a problem, at least in the specific case of the Route53
implementation as it always calls the ApplyChanges method, which in its
turn always retrieves all records in all zones.
This causes high pressure on Route53 APIs, for non-necessary actions.
By being able to filter all unmanaged records from the plan, we can
prevent from calling ApplyChanges when nothing has to be done and hence
prevent an unnecessary listing of records.
By doing so, the rate of API calls to AWS Route53 is expected to be
reduced by 2
* adds tests for shouldUpdateProviderSpecific
Signed-off-by: Raffaele Di Fazio <difazio.raffaele@gmail.com>
* move AWS health to where it belongs
Signed-off-by: Raffaele Di Fazio <difazio.raffaele@gmail.com>
* add test that breaks things
Signed-off-by: Raffaele Di Fazio <difazio.raffaele@gmail.com>
* adds adjustendpoints method
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* fix controller
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* actually pass the provider where needed
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* OMG goland do your go fmt thing
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* use registry as proxy
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* make linter happy
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* change AdjustEndpoints signature
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* fix typo
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* actually use adjusted endpoints
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* revert cloudflare change
Signed-off-by: Raffaele Di Fazio <raffo@github.com>
* Update provider/cloudflare/cloudflare.go
Co-authored-by: Nick Jüttner <nick@juni.io>
Co-authored-by: Nick Jüttner <nick@juni.io>
The controller will retrieve all the endpoints at the beginning of its
loop. When changes need to be applied, the provider may need to query
the endpoints again. Allow the provider to skip the queries if its data was
cached.
* Make suitableType() be Endpoint method
With this change it becomes possible to work with endpoint
of empty type in packages other than "provider". Also
it seems logical for a smart property getter without side effects
to be a method rather than a function in different package
* Make plan computation work correctly with multi-target domains
* fix drawing
* drop comments
* fix boilerplate header
* fix comment
* fix the bug with empty map
* rework registry to support random lables
* serialize->serializeLabel function rename
* golint for err variable naming
* add additional test
* add tests for current case where one resource can generate multiple endpoints
* make labels have its own type, add serialization as a method
* add comment for exported error
* use greater rather than not equal zero
* update changelog
* continue on controller loop error
* add logging in source
* use formatter on logChanges for endpoints
* fix log messages, log skipped records
* add logging in aws, uppercase the rest
* respect google dry run policy
* add ing/svc namespace/name on logging
* fix error logging on template failure
* fix bugs, propagate template error
* log if nothing is being updated, debug log skipped endpoints
* change zone-not-found logging order
* kickoff txt registry
* fix inmemory dns provider to include recordtype info for validation
* Merge master
* fix ununsed variable in inmemory provider
* add tests for records
* add test for no prefix name formatter
* implement apply changes with tests
* add flag to enable txt registry
* add txt registry to main
* improve sort testing
* filter out non-owned records
* NewEndpoint(...) requires record type
* use newendpoint in aws_test, fix tests
* change suitable type implementation
* fix the test for compatibility component
* change inmemory provider to include recordtype and use suitable type
* fix comments, CNAME should target hostname
* name mapper do not use pointer on struct
* txt prefix - just concatenate, remove spew, fix txt record label
* allow TXT records as result from dns provider
* add changelog
* fix tests
* TXT records need to be enclosed in double quotes
