Since Velero v1.12.0 and newer, backing up a non-existing namespace is now reporting as "Completed" instead of "Failure" this affects our failed alerts

[egondalia]

Velero Version 1.11.1 and older output STATUS worked ok, either a Failed or PartiallyFailed running below: (see results below)

1. Velero Versions =< 1.11.1 - Create backup of a non-existing NameSpace (status fails ok)

velero backup create mybackup1-11.1-and-older --include-namespaces 'nonexisting1ns' --include-resources '*' --include-cluster-resources=true

Results

velero get backup
NAME                                             STATUS                ERR   WARN   CREATED                         
mybackup1-11.1-and-older                         **PartiallyFailed**   1     0          2024-02-14 09:44:05 +0000 UTC   

###########################################################################################

Running the same command on Velero Version(s) greater than 1.12.0 (status now shows "Completed" but should show as failure because the namespace does not exist)

2 Velero Version above > 1.11.1 - Backup of a non-existing NameSpace (shows Completed should state failure BUG)

velero backup create mybackup1-12-0-and-above --include-namespaces 'nonexisting1ns'' --include-resources '*' --include-cluster-resources=true
Results

(Note the 2nd job should states a failure but shows up Completed, the Namespace does not exist but still completes)

velero get backup
NAME                                              STATUS            ERR   WARN   CREATED                         
mybackup1-11.1-and-older                      PartiallyFailed       1     0          2024-02-14 09:44:05 +0000 UTC   
mybackup1-12-0-and-above                       **Completed**         0     0          2024-02-14 10:03:51 +0000 UTC   

What did you expect to happen:
The second job should also have failed or reported as PartiallyFailed like previous versions of Velero did, any version above 1.11.1 is now outputs the Status as Completed instead of failed (note: The Namespace I am backing up does not exist so should continue to fail on newer version)

[ywk253100]

The change was introduced by #6320

[egondalia]

Can you please explain how #6320 is related to this the issue. Our issue is that previously a Backup would 'fail' if we created a backup job on a non-existing namespace but since 1.12.0 if we run a backup on a non-existing namespace it reports as 'Completed' instead of a 'fail' or 'partial fail'.

So if someone had removed a namespace but had a existing scheduled job to backup this namespace up, it would still continue to state 'completed' even though the namespace no longer exists. (since 1.12.0 we no longer get a failed alert via our prometheus)

[ywk253100]

#6320 removed the logic to check the existence of namespaces

[ywk253100]

Create a PR to fix this issue: #7569
After the PR merged, the backup will report a validation error if the specified namespaces don't exist