Multiple clusters backup

[Leo David]

Hello,
I see this great feature planned on the roadmap:
"Backup to one (physical) datastore from multiple Proxmox VE clusters, avoiding backup naming conflicts"

At this moment, I think that the only way to separate clusters with same vmid's would be to have them connected to different pbs datastores. Please correct me if I am wrong.

Does anybody have a rough ideea when this feature would be available ( even as a beta release ) ?

Thank you,

Leo

 

[Stefan_R]

No timeline yet, however, keep in mind that a 'datastore' is just that: a datastore in the PBS GUI.

It is certainly possible to back up several clusters to one PBS instance, if you add more then one datastore (can even be on the same physical storage, just use a subdirectory for each one).

The only thing you're losing is deduplication, since that works only within one datastore. And I suppose even that can be mitigated by using ZFS dedup on the storage level below, even if it might cost some performance.

 

[Leo David]

Thank you so much Stefan,
Indeed, having different directories really should help for the minute.
Thank you very much for hint !
Cheers,

Leo

 

[Oliver Albert]

yes different Directories work fine.
Backing up 2 Proxmox Clusters and even some machines as filelevel in a third directory. All on a ZFS-Volume.
I'm using my second backup-Machine as replica-Server.

Best Regards,
Oliver

 

[Leo David]

Hi,
I have just tried to configure two remote pbs synking to a central pbs and i am having some issues.
On the central pbs I only have one block drive ( one raid 1 array )
I have added the array as a disk ( without adding it as a datastore ) so the mountpoint looks:
device: /dev/disk/by-uuid/disk_id_number mounted on : /mnt/datastore/array-1

The I have created a couple of datastores as directories under:
/mnt/datastore/array-1/datastore-1
/mnt/datastore/array-1/datastore-2
/mnt/datastore/array-1/datastore-3
/mnt/datastore/array-1/datastore-4

I could do the sync from the two remote servers to central pbs datastore-1 and respectivelly datastore-2, but the issue comes about each datastore own usage.
They are all displaying the whole mountpoint usage, and not each datastore own usage.

I am aware that this might be expected being that they are all sharing the same mointpoint, but my question would be
1. Did I wrongly understood and configured multiple datastores placed on the same block device ?
2. Is there other way to mitigate this usage report ?

Thank you,

Leo

 

[Stefan_R]

Your idea is correct, but we currently only report usage data per mount point. Reading disk usage data for specific subdirectories usually requires iterating all contents and is thus very expensive, especially for a chunk store.

If you need the data, I'd recommend du or the more graphical ncdu.

 

[TwiX]

Hi,

We use the same datastore for all my pve clusters without any issue.

Just be aware of your VM/CT IDs ! But if you have like me a unique id per vm, no pb.

 

[Leo David]

Thank you so much for replies, at least I am now confident that I have configured as per your advices.
I works well for the minute, and it does the job.
Looking forward for next features to be introduced.
Nice team, nice work, nice job !
Cheers,

Leo