PBS with two CTs of the same ID on different stand-alone nodes - Is this OK?

[helojunkie]

So we have a 6-node prox cluster and several stand-alone proxmox servers (all 8.0.4) with PBS at 3.0-3). PBS is available on all nodes in our cluster and our stand-alone servers.

Well, today, something weird happened. I ran a snapshot backup of a CT on one of our standalone nodes to our PBS, and it appears to have assumed the identity of a CT on our cluster using the same naming convention and everything. I say "assumed the identity" because I have no other way to explain it. Both snapshots show up under ct/101 in my root namespace.

They are listed separately, and the comments for each match the correct snapshot, everything looks fine, I did a test restore on both, and it worked, I guess my question is if this is the expected way to do this or if am I messing something up using PBS in this manner? If I didn't have a comment on the snapshots I am not sure I would have known which snapshot was which since they are both under ct/101.

 

[Dunuin]

Best practice is to never use the same VMID twice to not run into these problems in the first place.
What I would do is splitting the datastore in multiple namespaces and then give each cluster or node it's own namespace. That way all data still gets deduplicated but it doesn't matter if multiple Nodes use the same VMIDs.