How to start a VM on another cluster if home server is not available

[Aron Dijkstra]

Hi,

I wonder how can we start a VM on another server when the server that has the configuration is stopped?
There is a cluster but HA is not available at the moment because we are setting up and testing fencing.

Using Migrate when the node with the configuration is down does not work....

With regards,

Aron

 

[fireon]

How many nodes to you have in your cluster?

 

[Aron Dijkstra]

3 nodes, we have everything for HA. But the question is if a node will not get HA, how can we migrate it if that server is down?

 

[fabian]

if the node is down for sure and all the disks are on shared storage, you can move the config file in /etc/pve/ from one node sub folder to another (manually, on the shell, with "mv"). but you must only do this when you are 100% sure that the node and VM are down!

 

[Aron Dijkstra]

Ok, thanks! that is what i needed
There is no refresh needed or a reboot to see the configuration in the other server GUI?

 

[fabian]

no, everything in /etc/pve is synced in a cluster.

 

[semira uthsala]

@fabian

Sorry for bringing this up on an old topic.

if the node is down for sure and all the disks are on shared storage, you can move the config file in /etc/pve/ from one node sub folder to another (manually, on the shell, with "mv"). but you must only do this when you are 100% sure that the node and VM are down!

After doing this and start VMs on another node on the cluster, How can I safely bring up the stopped/crashed server back?

I get that if I start the stopped server again it will corrupt the volumes of VMs that I already moved to another node manually. How to avoid that and bring back the stopped node? Is this possible to do without removing the failed node from the cluster?

Thanks

 

[ph0x]

If you moved the content in /etc/pve then the node will sync this directory upon reboot and realize that the VM has been moved. There should be no trouble.