Proxmox HA and Replication Questions

[rbeard.js]

Hello Everyone,

We were looking into the HA functions for proxmox but were having some issues where if we reboot one of our nodes, HA will kick in and move all the VMs over even though the reboot was intentional. Is there a way to get around this behavior? Even if we power down the VMs the issue still persists unfortunately and that doesnt seem right.

Is there a way to set the VMs to go back to their original server after an internet outage is fixed and the original host is back online automatically?

Also we replicate or VMs across the hosts nightly with snapshots. Is there a way to manually spin up those replicated images?
For example we have VM1 on server 1 and its having a windows issue. We will need to roll the server back to our last snapshot. In the meantime, it needs to be running. VM1 is replicated to server 2. I wouldnt want to migrate the VM as it will pull over the potential issues. How can I spin up that replicated VM on server 2 now that VM 1 on server 1 is shut down?

 

[fiona]

Hi,

Hello Everyone,

We were looking into the HA functions for proxmox but were having some issues where if we reboot one of our nodes, HA will kick in and move all the VMs over even though the reboot was intentional. Is there a way to get around this behavior? Even if we power down the VMs the issue still persists unfortunately and that doesnt seem right.

this is expected with the default shutdown policy, but you can use the freeze policy to avoid any migration/relocation.

Is there a way to set the VMs to go back to their original server after an internet outage is fixed and the original host is back online automatically?

You can create a HA group for those services and set the current node to a higher priority. When nofailback is disabled (default), the services will get migrated back to the higher-priority node.

Also we replicate or VMs across the hosts nightly with snapshots. Is there a way to manually spin up those replicated images?
For example we have VM1 on server 1 and its having a windows issue. We will need to roll the server back to our last snapshot. In the meantime, it needs to be running. VM1 is replicated to server 2. I wouldnt want to migrate the VM as it will pull over the potential issues. How can I spin up that replicated VM on server 2 now that VM 1 on server 1 is shut down?

I'm afraid that's not really possible without manually doing things. You could move the VM configuration manually to the other node, and the VM will start from the snapshot present on that node of course. But this is not officially supported and you might need to disable replication for the time being (replication will try to replicate the current state back to the other node automatically).

 

[rbeard.js]

Thank you so much for the feedback. Tested these options out and they are working like we hoped now!

 

[rbeard.js]

Hi,

this is expected with the default shutdown policy, but you can use the freeze policy to avoid any migration/relocation.


You can create a HA group for those services and set the current node to a higher priority. When nofailback is disabled (default), the services will get migrated back to the higher-priority node.


I'm afraid that's not really possible without manually doing things. You could move the VM configuration manually to the other node, and the VM will start from the snapshot present on that node of course. But this is not officially supported and you might need to disable replication for the time being (replication will try to replicate the current state back to the other node automatically).

One more question if you dont mind,
Both my servers and my Proxmox Backup Server are all setup on ZFS pools. If I wanted to download or copy my entire raw disk to use on either another machine or emulator, is there a way I could do that either from the backup server or the server itself?

 

[fiona]

From the Proxmox Backup Server, you can get the raw disk via e.g.

proxmox-backup-client restore vm/1000/2022-03-23T12:37:11Z drive-scsi0.img vm-1000-disk-0.raw --repository 192.168.20.180:datastore

which will restore the scsi0 drive from that VM's backup as the file vm-1000-disk-0.raw.

On Proxmox VE, there should be block device links in /dev/zvol/<pool>/vm-<ID>-disk-<N> which can be copied directly or via qemu-img convert while the VM is not running!

 

[rbeard.js]

From the Proxmox Backup Server, you can get the raw disk via e.g.

proxmox-backup-client restore vm/1000/2022-03-23T12:37:11Z drive-scsi0.img vm-1000-disk-0.raw --repository 192.168.20.180:datastore

which will restore the scsi0 drive from that VM's backup as the file vm-1000-disk-0.raw.

On Proxmox VE, there should be block device links in /dev/zvol/<pool>/vm-<ID>-disk-<N> which can be copied directly or via qemu-img convert while the VM is not running!

Thank you again!
Its my first time working with ZFS pools so its been a departure from what I would usually could do; dive into the system files and copy the disk in file explorer lol. This is a huge help