ProxMox 6 - 2 node cluster RAID 10 - how to migrate vm?

[rapture]

Hi,

thank you at all for any help....

We need to create a 2 nodes cluster. Every node is in RAID 10.

Is there a way to configure the ProxMox to migrate machines from a node to another whithout the third node?

Thank you at all.

 

[fiona]

Hi,
yes, you can create a two node cluster and migrate VMs between the two nodes without any special configuration.
Although you can't use the High Availability manager, which needs at least three nodes.

 

[rapture]

Ok thanks, but how?

Can we do that without a storage in common between the nodes?

I received this error:

root@virtual1:~# qm migrate 100 virtual2 --online
Requesting HA migration for VM 100 to node virtual2


root@virtual1:~# qm migrate 101 virtual2
2020-01-22 17:28:51 starting migration of VM 101 to node 'virtual2' (10.10.0.2)
2020-01-22 17:28:51 found local disk 'local-lvm:base-101-disk-0' (in current VM config)
2020-01-22 17:28:51 copying disk images
illegal name 'base-101-disk-0' - sould be 'vm-101-*'
command 'dd 'if=/dev/pve/base-101-disk-0' 'bs=64k'' failed: got signal 13
send/receive failed, cleaning up snapshot(s)..
2020-01-22 17:28:51 ERROR: Failed to sync data - command 'set -o pipefail && pvesm export local-lvm:base-101-disk-0 raw+size - -with-snapshots 0 | /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=virtual2' root@10.10.0.2 -- pvesm import local-lvm:base-101-disk-0 raw+size - -with-snapshots 0' failed: exit code 255
2020-01-22 17:28:51 aborting phase 1 - cleanup resources
2020-01-22 17:28:51 ERROR: found stale volume copy 'local-lvm:base-101-disk-0' on node 'virtual2'
2020-01-22 17:28:51 ERROR: migration aborted (duration 00:00:01): Failed to sync data - command 'set -o pipefail && pvesm export local-lvm:base-101-disk-0 raw+size - -with-snapshots 0 | /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=virtual2' root@10.10.0.2 -- pvesm import local-lvm:base-101-disk-0 raw+size - -with-snapshots 0' failed: exit code 255
migration aborted

What's wrong?

Thank you.

 

[fiona]

Ok thanks, but how?

Can we do that without a storage in common between the nodes?

I received this error:

root@virtual1:~# qm migrate 100 virtual2 --online
Requesting HA migration for VM 100 to node virtual2

Seems like the VM is HA managed. It turns out you can use HA with only two nodes as well, but it's not recommended. The migration of the machine should be started by HA after a bit. Is there nothing in your task log (the bottom panel in the GUI)? There should be an entry VM 100 - Migrate and one HA 100 - Migrate. By double clicking on that entry you should get the log.

root@virtual1:~# qm migrate 101 virtual2
2020-01-22 17:28:51 starting migration of VM 101 to node 'virtual2' (10.10.0.2)
2020-01-22 17:28:51 found local disk 'local-lvm:base-101-disk-0' (in current VM config)
2020-01-22 17:28:51 copying disk images
illegal name 'base-101-disk-0' - sould be 'vm-101-*'
command 'dd 'if=/dev/pve/base-101-disk-0' 'bs=64k'' failed: got signal 13
send/receive failed, cleaning up snapshot(s)..
2020-01-22 17:28:51 ERROR: Failed to sync data - command 'set -o pipefail && pvesm export local-lvm:base-101-disk-0 raw+size - -with-snapshots 0 | /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=virtual2' root@10.10.0.2 -- pvesm import local-lvm:base-101-disk-0 raw+size - -with-snapshots 0' failed: exit code 255
2020-01-22 17:28:51 aborting phase 1 - cleanup resources
2020-01-22 17:28:51 ERROR: found stale volume copy 'local-lvm:base-101-disk-0' on node 'virtual2'
2020-01-22 17:28:51 ERROR: migration aborted (duration 00:00:01): Failed to sync data - command 'set -o pipefail && pvesm export local-lvm:base-101-disk-0 raw+size - -with-snapshots 0 | /usr/bin/ssh -e none -o 'BatchMode=yes' -o 'HostKeyAlias=virtual2' root@10.10.0.2 -- pvesm import local-lvm:base-101-disk-0 raw+size - -with-snapshots 0' failed: exit code 255
migration aborted

What's wrong?

Thank you.

The second one fails, because it is a template and we currently do not support migration of templates. But we might in the future, here is the feature request. As a workaround you can create a full clone of the template, migrate the clone and then convert to template on the target node.

 

[rapture]

DRBD was the right way for me!
Thanks!