Closed a7i closed 1 month ago
/cc @jwcesign
would you be open to reviewing? seems like a regression from this
:warning: Please install the to ensure uploads and comments are reliably processed by Codecov.
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 28.26%. Comparing base (
4ba18c1
) to head (2530ab9
).
:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
/retest
Hi @a7i, thanks for your feedback. I'm sorry for the disruption to your business.
This also reminds me that we do not have E2E capabilities to maintain related capabilities. Do you think we can design some E2E capabilities to maintain the logic?
Hi @a7i, thanks for your feedback. I'm sorry for the disruption to your business.
This also reminds me that we do not have E2E capabilities to maintain related capabilities. Do you think we can design some E2E capabilities to maintain the logic?
all good! we're happy that we can make small contributions to this project.
Happy to explore this and submit a few e2e tests in a separate PR (so we can cover CRB and RB)
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: XiShanYongYe-Chang
The full list of commands accepted by this bot can be found here.
The pull request process is described here
What type of PR is this? /kind bug
What this PR does / why we need it:
We've been having a lot of issues with karmada deleting cluster-level resources such as ClusterRole and ClusterRoleBinding.
After debugging, we realized that the label
clusterresourcebinding.karmada.io/permanent-id
was missing from those ClusterResourceBinding and that the webhook is responsible for populating that. Given that this was empty, all resource propagated to the member cluster also had an empty value for the id (i.e.clusterresourcebinding.karmada.io/permanent-id: ""
) which is then identified as orphaned work and is deleted in the cluster.This must have broke after the "63 char limit issue" because all of our CRBs prior to that release are fine and do not cause orphan work issues:
(note 79 days ago is good, but all recent ones are not)
Which issue(s) this PR fixes: Fixes #
Special notes for your reviewer:
Does this PR introduce a user-facing change?: