* init release branch
* init 1.20 nightly tests
* drop 1.17 nightly tests for new release cycle
* drop 1.17 from test matrix
* Update nightly-test-integrations-1.20.x.yml
In some environments, the script will not fail despite SKIP_CHECK_BRANCH
being unset, leading to the script explicitly skipping CI when it should
fail fast.
Prevent this by explicitly checking for the env var.
* explicitly add content-type anywhere possible and add middleware to set and warn
* added tests, fixed typo
* clean up unused constants
* changelog
* fix call order in middleware
* remove v2 tenancy, catalog, and mesh
- Inline the v2tenancy experiment to false
- Inline the resource-apis experiment to false
- Inline the hcp-v2-resource-apis experiment to false
- Remove ACL policy templates and rule language changes related to
workload identities (a v2-only concept) (e.g. identity and
identity_prefix)
- Update the gRPC endpoint used by consul-dataplane to no longer respond
specially for v2
- Remove stray v2 references scattered throughout the DNS v1.5 newer
implementation.
* changelog
* go mod tidy on consul containers
* lint fixes from ENT
---------
Co-authored-by: John Murret <john.murret@hashicorp.com>
Correctly set the the version of Consul built by the `dev-build` job,
which is then copied into the Consul dev image used in integration
tests.
This was causing failures starting sidecar proxies via `consul connect
envoy` due to a mismatch between the (incorrect) Consul binary's
supported Envoy versions and the (correct) Envoy version under test.
Also add debug log uploads to each int test so we can more easily
diagnose this sort of failure in the future, as it was entirely hidden
in test output.
* Set replication metric to 0 when losing leadership
* Fix replication metrics on replication.go also
---------
Co-authored-by: sarahalsmiller <100602640+sarahalsmiller@users.noreply.github.com>
* run integration tests on push in main and release/*
* Update .github/workflows/test-integrations.yml
Co-authored-by: Michael Zalimeni <michael.zalimeni@hashicorp.com>
---------
Co-authored-by: Michael Zalimeni <michael.zalimeni@hashicorp.com>
k8s.gcr.io has been migrated to registry.k8s.io for several years now,
and the old registry is being shut down, causing image pull failures.
Update to target the new registry when pulling the pause image used in
Envoy integration tests.
* add build support script to print out the submodule versions required in other submodules
* update help and usage. exclude current submodule in output.
* Add checks to CI to ensure that generated golden files for xds tests are up to date
* fix file permissions
* debugging
* more debugging
* more debugging
* more debugging
* more debugging
* I can't type
* this might be correct
* removing debug prints
* get rid of unused column
* get rid of duplicate section now that deletion of unused column makes the section duplicate..
* explicit set protocol rathern than infer it in checkDNSService
* explicit have attribute for whether to set EDNS0 in the test cases rathern than infer it in checkDNSService
* now modify so that test agents are only created for each unique configuration which is based on the a_record_limit.
* Fix TestDNS_ServiceLookup_AnswerLimits so that it only creates test agents the minimal amount of time. (#21609)
Fix TestDNS_ServiceLookup_AnswerLimits so that it only creates test agents the minimal amount of time
Remove note that OpenShift 4.16 is not yet available, now that it's been released.
It will be added to the matrix in a future update once we've tested compatibility across eligible `consul-k8s` versions.
Dhia Ayachi