Upload progress monitoring for plugins after backup completion

[dsu-igeek]

Describe the problem/challenge you have

Volume snapshotter plug-in are used by Velero to take snapshots of persistent volume contents.
Depending on the underlying storage system, those snapshots may be available to use immediately,
they may be uploaded to stable storage internally by the plug-in or they may need to be uploaded after
the snapshot has been taken. We would like for Velero to continue on to the next part of the backup as quickly
as possible but we would also like the backup to not be marked as complete until it is a usable backup. We'd also
eventually like to bring the control of upload under the control of Velero and allow the user to make decisions
about the ultimate destination of backup data independent of the storage system they're using.

Describe the solution you'd like
Enable monitoring of uploads that are in-progress without blocking backups from starting.

Vote on this issue!

This is an invitation to the Velero community to vote on issues, you can see the project's top voted issues listed here.
Use the "reaction smiley face" up to the right of this comment to vote.

• 👍 for "The project would be better with this feature added"
• 👎 for "This feature will not enhance the project in a meaningful way"

[reasonerjt]

Moving this to icebox as we have decided to choose the alternative approach to track the progress: #5108

[arianitu]

@reasonerjt I just want to confirm that issue is not resolved in all released versions of velero. There are a lot of tasks that are closed that run into this particular issue (specifically restores failing because snapshots are not ready yet.)

We have patched our own version of velero that handles this case, I want to confirm that we should continue to maintain our own patched version until this is fixed via #5108?

Again, I want to mention for others that this issue may not be completely relevant if you're using CSI snapshots because they handle this case, here is my comment from a few years ago:

If anyone is curious, we talked to the Google folks managing the gke-csi diver and the CSI driver will handle this case for you. This may be relevant for you nrb since I don't think velero needs to wait for snapshots to be ready, so I guess you guys do not need to do anything.
Basically, the PVC provisioner will retry forever until the VolumeSnapshot is readyToUse. So in theory, the EBS and GKE volume snapshotter bugs are no longer relevant if you use the CSI snapshot.

I am unsure if CSI snapshots support restoring to another namespace, but our hope is one day to move to CSI snapshots to solve this issue.

Cheers,

[jglick]

Does https://github.com/vmware-tanzu/velero/blob/main/design/Implemented/general-progress-monitoring.md address this issue?

[blackpiglet]

@jglick
Yeah. This design was already implemented in Velero v1.11.0.

[sseago]

@blackpiglet @jglick Just a note, we've not yet implemented this for volumesnapshotter plugins (native aws, gcp, azure snapshots), but it is available for BackupItemAction/RestoreItemAction plugins as of 1.11, and will be used for the CSI plugin (and the new datamover) in 1.12