PVE 4.2: force migration of vm from a dead node to another alive node

[baudinpr]

Hi all,

just installed 2 PVE 4.2 servers in 2 node cluster configuration.
VM data is shared via DRBD9. All is working well.

Now I'm trying to do some tests...

I have just tried to shutdown a node and force migration of a VM running on that node to the other node.
On the running node, I tried:
# mv /etc/pve/nodes/DOWN-NODE-NAME/qemu-server/VMID.conf /etc/pve/nodes/UP-NODE-NAME/qemu-server/
but the /etc/pve filesystem is read only...

What I have to do ?

Many Thanks
Piero
PS both nodes have PVE community subscription active

 

[dietmar]

Your run a two node cluster, so you loose quorum if one node is down. As workaround, you can set
expected votes to 1 using:

# pvecm expected 1

Note: Only do that when the other node is really down!

 

[baudinpr]

Thanks Dietmar,

how to return to original config after forcing expected votes to 1 ?
Tried:
# pvecm expected 2
but:
Unable to set expected votes: CS_ERR_INVALID_PARAM

Piero

 

[dietmar]

simply start the second node

 

[baudinpr]

Ok, it worked!

So, I think I have to add all the VM to the HA Resources on datacenter, right ?

 

[baudinpr]

Yes!
It works great!

 

[dietmar]

So, I think I have to add all the VM to the HA Resources on datacenter, right ?

HA needs at least 3 nodes...