[SOLVED] PVE/Ceph: Strategy for disaster recovery backups of VMs

[sherminator]

Hi there,

we just started with our three node PVE/Ceph cluster - so far so good.
Now I'm looking for the best way for "disaster recovery style" backups of our Windows and Linux VMs.

At the moment we do backups like this:
For file based backups we use Bareos which works well. For (weekly) disaster recovery backups we use Windows system backup and Relax and Recover.
Regarding disaster recovery I like the idea of doing the same kind of backup with our Windows and Linux guests, so I took a look at Proxmox backup. Looks good, but:
1. Is it not possible to choose certain disks of a VM? We have large databases and file servers, and I just like to backup the system disks of them, because the data is backuped another way and would be too large.
2. During Proxmox backup (snapshots, LZO compressed to an NFS share) I noticed that some VMs were unreachable for some seconds or (in two cases) even 1-2 minutes. Is that as expected, or should I search for (configuration) mistakes?

Does somebody have experience with cv4pve-barc (https://github.com/Corsinvest/cv4pve-barc)?
What do you suggest as a disaster recovery backup strategy which fits our needs?

Thanks a lot and many greets
Stephan

 

[sherminator]

or in fewer words:
I would love to do weekly backups of the system disks of all VMs (snapshot, LZO compressed, to an NFS share). High level of consistency (as Proxmox backup does via qemu guest agent) would be fine, but I can live without it, too.
Can vzdump do this?
Can cv4pve-barc do this (and is it recommended)?
Other ways?

Thanks and greets
Stephan

 

[sherminator]

Update: Proxmox Backup is able to do what we want: You can go to VM -> Hardware -> Hard Disk -> Edit -> Advanced -> No Backup.

Greets
Stephan