2 node cluster vm migration between local storage issues

[FishCorp]

Hello Forum!

I'm trying to do something that i thought was a fairly simple job, but apparently not

I have a two node cluster:

Server1:
Disk1: System+VMStorage disk (450GB ZFS) listed as local + local-zfs under the node storage.

Server2:
Disk1: System NVME disk (120GB ZFS) listed as local + local-zfs under the node storage.
Disk2: VM Storage disk (700~ GB ZFS) listed as 'vmstore'

The VM i'm trying to migrate is on Server 2, stored on the 'vmstore' drive - and i can find no combination of migration-settings, online / offline that works. I've tried setting the target of the migration to the local-zfs drive on Server1 but the migration job fails.

Can anybody offer a solution (besides offline full back + restore)? Eg. could i rename the local-zfs pool on Server 1 to 'vmstore' somehow?

Best regards,
Rune / FishCorp

 

[VoIP-Ninja]

Hello,

for online-migration / live-migration with zfs your storage-pools need to be named same. E.g. you can only live-migrate from server1 / local-zfs to server2 / local-zfs and vice versa. From vmstore ( server2 ) to local-zfs ( server1 ) you have to shutdown the vm, then migrate.

greetings

EDIT: shutdown vm ( as server could be misinterpreted )

 

[FishCorp]

Hi VoIP-Ninja! Thank you for responding.

I'm a little wiser this morning, than i was last night when i wrote the post.

When migrating with the server running, i have the option in the interface to specify a storage-target on the remote node - but when the VM is powered off this "field" in the interface disappears! But - it doesn't work anyway . The job fails when migrating from 'vmstore' > 'local-zfs'.

I then moved the test-vm disk to local-zfs on the source server, and then the migration works as expected back and forth!

However: when i tested migrating from server1 (stored in 'local-zfs') to server2 with 'vmstore' set as a target - the job completes, but ends up on local-zfs on the remote node, despite having specified a different target .. so it would seem like i have run into this bug:

https://bugzilla.proxmox.com/show_bug.cgi?id=1804

Can anybody confirm this?

Best,
Rune