7979 fix

[blackpiglet]

Thank you for contributing to Velero!

Please add a summary of your change

Does your change fix a particular issue?

Fixes #7979

Please indicate you've done the following:

• Accepted the DCO. Commits without the DCO will delay acceptance.
• Created a changelog file (make new-changelog) or comment /kind changelog-not-required on this PR.
• Updated the corresponding documentation in site/content/docs/main.

[codecov]

Codecov Report

Attention: Patch coverage is 76.10922% with 70 lines in your changes missing coverage. Please review.

Project coverage is 59.57%. Comparing base (e64806a) to head (f98fca3).

Files with missing lines	Patch %	Lines
...delete/actions/csi/volumesnapshotcontent_action.go	62.00%	17 Missing and 2 partials ⚠️
pkg/util/csi/volume_snapshot.go	44.11%	15 Missing and 4 partials ⚠️
...estore/actions/csi/volumesnapshotcontent_action.go	65.11%	11 Missing and 4 partials ⚠️
pkg/restore/actions/csi/volumesnapshot_action.go	84.12%	8 Missing and 2 partials ⚠️
pkg/cmd/server/plugin/plugin.go	0.00%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #8684      +/-   ##
==========================================
+ Coverage   59.48%   59.57%   +0.08%     
==========================================
  Files         371      370       -1     
  Lines       40191    40199       +8     
==========================================
+ Hits        23909    23947      +38     
+ Misses      14789    14758      -31     
- Partials     1493     1494       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[blackpiglet]

@anshulahuja98
Please also take a look. I made some changes in the design because I found that restore needs a way to record the VSs and VSCs generated in the restore.
As a result, I added an async operation in the VolumeSnapshot RestoreItemAction to wait till the VSC's snapshot handle is created.

Although the async operation doesn't wait till VS and VSC become ReadyToUse, it is still different from what we have agreed.
Please check whether that's OK for your scenario.

[anshulahuja98]

As discussed over chat, I have concern with deletion of VS / VSC in finalizing.
Our eventual targe is that the PVC being restored from VS gets restored properly.
Now here if we delete the VS before the PVC is fully hydrated.
Consider 2 scenarios - If PVC is waiting for first consumer and there is no pod mounting it, and vS gets deleted, the VS is useless

Another scenarios is where CSI driver is expected to take time to hydrate the PVC from VS, in those case also we can't delete VS inflight.

[blackpiglet]

As discussed over chat, I have concern with deletion of VS / VSC in finalizing. Our eventual targe is that the PVC being restored from VS gets restored properly. Now here if we delete the VS before the PVC is fully hydrated. Consider 2 scenarios - If PVC is waiting for first consumer and there is no pod mounting it, and vS gets deleted, the VS is useless

Another scenarios is where CSI driver is expected to take time to hydrate the PVC from VS, in those case also we can't delete VS inflight.

It's a reasonable consideration. I will update the implementation.
Restore will not delete the VolumeSnapshotContent and VolumeSnapshot.