Hello,
I would like to back up our physical nodes which are in a cluster (PVE 7.2). No VMs/containers etc to be taken into consideration we already have backup solutions for those, they are not on local storage of any nodes.
At the moment when a node's boot storage dies or has any issues we have to reinstall the node. Reinstalling involves a very manual process of removing all traces of the "old" node from the cluster by editing config files and running commands, to then rejoin the "new" (same machine, fresh install) node.
We'd like to just be able to image the nodes boot storage, we already have veeam in our environment so this would be extremely easy with their linux agents. We have the veeam method fairly streamlined and efficient with pxe boot for recovery media, ipmi access etc so we're talking about 5 minutes of human intervention for a complete restore, excluding whatever machine time for data transfer but proxmox boot disks are tiny so it won't be much.
This works for non-clustered nodes so we know the concept itself is fine.
We just need to validate if there are any considerations for clustered instances, we're especially worried/unsure about the cluster file system shared by all nodes. Would restoring 1 node that is not in sync (backup would be daily so 24 - 48h out of sync max) cause issues or would the remaining nodes that have quorum just tell the restored node what data it should have and push the correct data? We think this would just be similar to a node being shut down for a while then powered on, but we want to be sure.
Other details:
I would like to back up our physical nodes which are in a cluster (PVE 7.2). No VMs/containers etc to be taken into consideration we already have backup solutions for those, they are not on local storage of any nodes.
At the moment when a node's boot storage dies or has any issues we have to reinstall the node. Reinstalling involves a very manual process of removing all traces of the "old" node from the cluster by editing config files and running commands, to then rejoin the "new" (same machine, fresh install) node.
We'd like to just be able to image the nodes boot storage, we already have veeam in our environment so this would be extremely easy with their linux agents. We have the veeam method fairly streamlined and efficient with pxe boot for recovery media, ipmi access etc so we're talking about 5 minutes of human intervention for a complete restore, excluding whatever machine time for data transfer but proxmox boot disks are tiny so it won't be much.
This works for non-clustered nodes so we know the concept itself is fine.
We just need to validate if there are any considerations for clustered instances, we're especially worried/unsure about the cluster file system shared by all nodes. Would restoring 1 node that is not in sync (backup would be daily so 24 - 48h out of sync max) cause issues or would the remaining nodes that have quorum just tell the restored node what data it should have and push the correct data? We think this would just be similar to a node being shut down for a while then powered on, but we want to be sure.
Other details:
- PVE 7.2
- all VMs and other production data is external shared storage, no local data
- VMs are HA so when a node dies they are moved to other hosts
- we would back up and restore nodes as needed using disk imaging (veeam in this case but same concept as any other image method)