New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix runtime admission flaky test due to race condition #91315
Conversation
Hi @jherrera123. Thanks for your PR. I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with Once the patch is verified, the new status will be reflected by the I understand the commands that are listed here. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
/sig node |
/retest |
see the verify errors that need fixing in https://prow.k8s.io/view/gcs/kubernetes-jenkins/pr-logs/pull/91315/pull-kubernetes-verify/1263304978068410368 |
/retest |
/lgtm |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: jherrera123, liggitt The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
/retest |
What type of PR is this?
/kind flake
What this PR does / why we need it:
This change fixes a problem in the
RuntimeClass
admission plugin rejecting pods with aRuntimeClass not found
message caused by a race condition.Which issue(s) this PR fixes:
Fixes #90653
Special notes for your reviewer:
RuntimeClass admission plugin uses a cache to determine if a pod should be admitted. This cache is updated when a RuntimeClass is created in an informer thread.
The cache updating thread runs in a different thread than the admission validation. If a RuntimeClass is created, and a pod with a RuntimeClass spec is immediately created afterwards, there's a window of time where the cache hasn't been updated yet, so the plugin just throws a
RuntimeClass not found
message and error out.With this change, if the cache lookup fails, we do a live lookup instead of waiting for the cache to be updated.
Unit tests were added that cover the exported methods
ValidationInitialization
andAdmit
, which are used in the modified code.I was also able to reproduce this problem consistently by adding a delay in the informer updating thread, in method
sharedIndexInformer.HandleDeltas
, and confirmed this change fixes the reported problem.Does this PR introduce a user-facing change?:
/cc @liggitt