Question about using namespaces in Proxmox Backup Server with multiple PVE clusters

[pvpaulo]

Hello, good afternoon everyone.

Could you please help me clarify a question?

I currently have a Proxmox Backup Server (PBS) properly configured, with a datastore named BKP.
I have one Proxmox VE (PVE) cluster that is already storing its backups in this BKP datastore, using the default namespace (root).

I am now in the process of creating a second PVE cluster, and I would like to use the same BKP datastore on the PBS to store its backups.

In this context, I have a few questions:

1. If I mount the BKP datastore on the second PVE cluster using the default (root) namespace, could this cause any issues or even corrupt the existing backups?
2. If I create a new namespace for the second PVE cluster, would there be any problems considering that the first cluster will continue writing backups to the root namespace?
3. When creating a new namespace, it is necessary to select the parent namespace (root). Could this hierarchy cause any impact or risk to the existing backups stored in the default namespace?
4. Is it safe to keep:
   
   • Cluster 01 writing backups to the root namespace, and
   • Cluster 02 writing backups to a new namespace created under root?

Or would the ideal scenario be:

• To create two separate namespaces,
• And assign one exclusive namespace to each cluster,
  considering that there are already backups stored in the default (root) namespace and that I cannot afford to lose them?

Thank you very much in advance for your help and guidance.

Best regards,

 

[Johannes S]

Hello, good afternoon everyone.

Could you please help me clarify a question?

I currently have a Proxmox Backup Server (PBS) properly configured, with a datastore named BKP.
I have one Proxmox VE (PVE) cluster that is already storing its backups in this BKP datastore, using the default namespace (root).

I am now in the process of creating a second PVE cluster, and I would like to use the same BKP datastore on the PBS to store its backups.

In this context, I have a few questions:

1. If I mount the BKP datastore on the second PVE cluster using the default (root) namespace, could this cause any issues or even corrupt the existing backups?

This might happen, if a vm or lxc shares it's numerical id with the ones in the default namespace if the new backups are written to this namespace as well. Although backups usually don't overwrite each other they might get messed up and a prune job or manual removal might remove backups you still want to keep.

1. If I create a new namespace for the second PVE cluster, would there be any problems considering that the first cluster will continue writing backups to the root namespace?

This should work but imho would be a little bit cluttered.

1. Is it safe to keep:
   • Cluster 01 writing backups to the root namespace, and
   • Cluster 02 writing backups to a new namespace created under root?

Or would the ideal scenario be:

• To create two separate namespaces,
• And assign one exclusive namespace to each cluster,
  considering that there are already backups stored in the default (root) namespace and that I cannot afford to lose them?

Both things should work, but I still would go with two seperate namespaces and change the existing backup job to the new namespace for cluster01. You could even create a sync job to sync the backups from root to the new namespace, you will just need to exclude the new namespace from the source (should be possible with a filter) to avoid a loop.

It would also be a good idea to create a dedicated API token for each namespace to limit the potential damage done by a bad actor.

Seperating makes it also easier to have different sync, prune etc schedules for each namespace if you have need for that.