Group filter

jlauro

Active Member
Feb 10, 2024
160
50
28
I am having trouble understanding how to configure the group filter in PBS. It appears to mainly be based on vmid, which is not very useful as there are multiple namespaces and so the id isn't unique, and also not maintainable. Can it be filtered based on vm name? What about originating pool or original labels? This is mainly in the 3rd PBS that receives copies from my main two PBS servers. It's on the other side of the country and can't keep up. I need a way to specify based on the vm configuration and not directly on the remote PBS server besides for initial setup of the pull PBS from PBS job. Suggestions?
 
  • Like
Reactions: rahman
Hi,
AFAIK, it's not currently possible to limit a group filter to a certain namespace for a job operating on multiple namespaces. For now, you probably need to create a job per namespace to work around this limitation.

Regarding the configuration-based filters, PBS is not currently designed to know about any details inside guest configurations. It would also mean reading and parsing all configurations each time, which is very expensive! Also, configurations and even VM names are not fixed for a group, but can change with each backup.

An idea could be to introduce tags (or similar) for a group (could then be passed along by the client). Currently, there already is the possibility to set a comment for a backup group. A filter based on the comment could be introduced.

Feel free to open a feature request linking back to this thread (after checking a similar one doesn't exist yet): https://bugzilla.proxmox.com/
 
https://bugzilla.proxmox.com/show_bug.cgi?id=3586 seemed to be the closest, so I added a comment and linked back.
VM names change, but we do have a naming convention for our servers and the names do show in the comments for the job. A regex filter on the comment to exclude all test vms would be very useful.
I'll look into setting up additional name spaces as a work around for now. That is a lot more configuration per cluster though, including creating multiple storage connections back to the same pbs server and multiple backup jobs.