test(e2e): validate that pulling from an acr registry attached to aks via --attach-acr with karpenter nodes works #369
Conversation
Pull Request Test Coverage Report for Build 9373575828Warning: This coverage report may be inaccurate.This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.
Details
💛 - Coveralls |
| var _ = AfterEach(func() { env.Cleanup() }) | ||
| var _ = AfterEach(func() { env.AfterEach() }) | ||
|
|
||
| var _ = Describe("Acr", func() { |
There was a problem hiding this comment.
I created an ACR suite rather than just adding to the existing integration suite because artifact streaming will require a suite for ACR behavior here as well.
|
I would recommend running it against the failing case (e.g., undoing #220) as well to validate it. |
| @@ -43,7 +43,7 @@ jobs: | |||
| strategy: | |||
| fail-fast: false | |||
| matrix: | |||
| suite: [Nonbehavioral, Utilization, GPU, Drift, Integration, NodeClaim, Chaos] | |||
| suite: [Nonbehavioral, Utilization, GPU, Drift, Integration, NodeClaim, Chaos, Acr] | |||
There was a problem hiding this comment.
Maybe AcrIntegration might be a better name? Not a strong opinion though.
But another thing to consider is that making it ACR might be more consistent with GPU one we have to the left.
There was a problem hiding this comment.
Yeah seems like a bit much
feat: checking in the acr suite test: adding steps for checking node and pod readiness refactor: precreating image name in before each fix: importing on same line chore: adding makefile helpers + pulling acr name from the environment refactor(ci): lint
Bryce-Soghigian
force-pushed
the
bsoghigian/e2e-acr-pull-clean
branch
from
959ce49 to
00f9d43
Compare
|
gonna rewrite this on the azure repo rather than my fork. |
Fixes #
Description
As part of the efforts to fix 1.30 bootstrap we removed a flag that removes functionality for out of tree providers. One of the side affects of this is that ACR Pull will not work in the same way. We needed a way to validate this image pull works in a reproducible and consistent way.
This test suite is that. Currently it has to be ran locally to target specific k8s versions so a nice followup at lower priority would be to add the ability to run our e2es against a specific k8s version.
How was this change tested?
Does this change impact docs?
Release Note