VM Migration in node failover

[Ami10]

Hello Everyone,

I have deployed Proxmox Cluster with Ceph Cluster in 3 nodes.

node1.omega.local
node2.omega.local
node3.omega.local

vm-100 is running on node2 with HA configuration. QEMU agent and balloon driver is installed on the vm-100.

When I migrate vm from node2 to node3, it is live migration, but if the node2 is fail the vm migration is offline mode (VM is restarting in another node).

I think this is because the RAM data of the running VM is not provisioned in another node.

So Is there a way to setup live migration for node failover.

Thanks and regards.

 

[t.lamprecht]

node2 is fail the vm migration is offline mode (VM is restarting in another node).

If a node is suddenly failing, i.e., got bricked, local power loss or something the like one cannot live migrate anymore, as the source from which to live migrate from isn't there anymore. So if a node goes real down it will always be a new start of the VM/CT once recovered to a working node.

If you mean on planned regular node shutdown (e.g., for maintenance), then you set the HA Shutdown policy to "migrate" in the Webinterface (Datacenter -> Options -> HA Settings). For shutdowns and reboots the service should then get live migrate away before the node powers off.

 

[Ami10]

Thanks for your reply.
Yes I am testing it by power off the server. I supposed it will live migration. VM will resume in another node from last state.

 

[t.lamprecht]

As said, if you do a software triggered power off then:

If you mean on planned regular node shutdown (e.g., for maintenance), then you set the HA Shutdown policy to "migrate" in the Webinterface (Datacenter -> Options -> HA Settings). For shutdowns and reboots the service should then get live migrate away before the node powers off.

Works since Proxmox VE 6.1