Remove kubevirt/kubevirt e2e workarounds once their upstream PRs merge

[kaovilai]

#2404's kdm e2e suite carries three workarounds for still-open kubevirt/kubevirt bugs. Each should be removed once its blocking PR merges and a release picks it up.

(migtools/kubevirt-datamover-controller#207, migtools/kubevirt-datamover-controller#208, and migtools/kubevirt-datamover-controller#212 are expected to merge shortly and are intentionally not tracked here.)

Workarounds to remove

1. Stuck VMB finalizer blocks namespace deletion forever (VirtualMachineBackup finalizer is never removed when its VirtualMachineBackupTracker no longer exists, blocking namespace deletion forever kubevirt/kubevirt#18724)

   • Workaround: tests/e2e/lib/virt_helpers.go's ClearStuckVMBFinalizers (line 1674) / IsNamespaceDeletedClearingStuckVMBFinalizers (line 1715), called from tests/e2e/backup_restore_suite_test.go's deleteNamespace (line 421)
   • Remove once: kubevirt/kubevirt#18725 merges. kubevirt/kubevirt#18289 alone is not sufficient -- it reduces how often the bug fires but leaves a timing/version-skew gap (confirmed by cross-check), so it complements Release the VMBackup finalizer when its BackupTracker is gone kubevirt/kubevirt#18725 rather than replacing it.

2. VirtualMachineBackup status freeze during target PVC attach (CNV-85377/CNV-89684)

   • Workaround: tests/e2e/lib/virt_helpers.go's NudgeVmiToTriggerResync (line 1054) and VMBHasNoConditions (line 1646), called from tests/e2e/virt_backup_restore_suite_test.go (lines 476, 482, 510)
   • Remove once: kubevirt/kubevirt#18949 merges.

3. Stale reconcile can mark an already-completed VirtualMachineBackup Failed with SourceLost

   • Workaround: tests/e2e/lib/apps.go's GetNamespaceEventMessages (line 386), called from tests/e2e/virt_backup_restore_suite_test.go (lines 507, 530)
   • Remove once: kubevirt/kubevirt#18957 merges.

Not tracked here (already a permanent improvement, not removable scaffolding)

#2404's hyperConvergedGVR() (preferring HCO's v1 API over v1beta1) stays useful even after kubevirt/hyperconverged-cluster-operator#4552 merges, so it's not a candidate for removal.

Note

Responses generated with Claude