fix uncertain cache

[huww98]

What type of PR is this?
/kind bug

Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line:

/kind api-change
/kind bug
/kind cleanup
/kind design
/kind documentation
/kind failing-test
/kind feature
/kind flake

What this PR does / why we need it:

fix wrong error is used when checking final error.

map is not safe for concurrent read/write from multiple worker. Also fixed that.

Added related tests.

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a user-facing change?:

Fixed uncertain state

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: huww98
Once this PR has been reviewed and has the lgtm label, please assign xing-yang for approval. For more information see the Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[k8s-ci-robot]

Hi @huww98. Thanks for your PR.

I'm waiting for a kubernetes-csi member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

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-sigs/prow repository.

[huww98]

We don't need to add them manually, as informer will read them from kubeClient. ctrl.init(ctx) will wait for cache sync.

[huww98]

Already done by caller, no need to do it again.

[huww98]

After a PVC is marked as completed, it is re-queued. In the next cycle, we may still read out the PVC with InProgress status. Then we should have conflict when trying to mark it InProgress again. However, fake client does not supports resourceVersion. So when we pass an empty patch to it, it returns the latest PVC, with ModifyVolumeStatus == nil, which will make markControllerModifyVolumeCompleted panic.

Should we somehow make metadata.generation work for PVC? So that we can cache the already synced generation+VACName, and will know the PVC in the cache is outdated, and just skip that cycle.

[sunnylovestiramisu]

I am trying to understand how it works, the metadata.generation field in a Kubernetes object is a sequence number that is only incremented when there is a change to the object's spec.

But ModifyVolumeStatus is in Status.ModifyVolumeStatus, therefore, caching the generation number would be ineffective? It wouldn't tell you if the status has been updated, so you would still be working with stale data?

Also Hemant raised an issue: #514 that we will fix before GA.

[huww98]

My plan is: we maintain a global sync.Map which maps from object UID to the generation that is has been fully reconciled. We add to this map when ControllerModifyVolume finishes with OK and specVacName matches targetVACName. We clear it when the object is deleted. When the next cycle starts, we just skip if the generation matches. This should ensure we will not sync the same VAC at the same generation twice.

Anyway, I think this is minor and should not block this PR.

Also Hemant raised an issue: #514 that we will fix before GA.

Yes, I think this PR should fix that one.

[gnufied]

/ok-to-test

[huww98]

Fixed the test. The test is broken during rebase, it is originally developed based on #513

[sunnylovestiramisu]

For the uncertain cache fix and change it to use sync.Map

/lgtm