external-snapshotter does not retry removal of VolumeSnapshotContent finalizers

[jsafrane]

What happened:
On an API server hiccup, the external-snapshotter was not able to remove its finalizer from VolumeSnapshotContent and got a timeout. Such hiccups happen all the time, however, the snapshotter did not retry with exp. backoff.

What you expected to happen:
The snapshotter re-tries with exponential backoff.

How to reproduce it:
I can't reproduce it on demand, it needs careful timing of the API server error. This small commit shows how removeContentFinalizer() call error causes the snapshotter not to retry: jsafrane@3da05bb

With that commit, I can see in the snapshotter logs just once:

I0515 16:52:17.267236       1 snapshot_controller.go:59] synchronizing VolumeSnapshotContent[snapcontent-2cee4522-4ae0-40d4-9718-4b51c1e261f8]
I0515 16:52:17.267260       1 snapshot_controller.go:631] Check if VolumeSnapshotContent[snapcontent-2cee4522-4ae0-40d4-9718-4b51c1e261f8] should be deleted.
I0515 16:52:17.267278       1 snapshot_controller.go:62] VolumeSnapshotContent[snapcontent-2cee4522-4ae0-40d4-9718-4b51c1e261f8]: the policy is Delete
E0515 16:52:17.267344       1 snapshot_controller_base.go:361] could not sync content "snapcontent-2cee4522-4ae0-40d4-9718-4b51c1e261f8": snapshot controller failed to update snapcontent-2cee4522-4ae0-40d4-9718-4b51c1e261f8 on API server: mock finalizer removal error

I want the snapshotter to retry periodically.

Environment:

• Driver version: csi-driver-hostpath
• csi-snapshotter: v8.2.0

[yati1998]

Hi @jsafrane , I see similiar issue already raised: #1282
The volumeSnapshotContent is not requeued

[jsafrane]

I think its a different issue. In #1282 re-queues a snapshot / snapshot content after 5 minutes and that may be too much in some cases. I don't see my snapshot content requeued ever.