Offline migration fails to move disk image

[n3n5n1]

Hello,

we've replaced one of the nodes in our cluster and now we can't do offline migration. Migration finishes in 10 seconds, GUI reports task results as successful ("TASK OK"), the VM in question is moved to a different node, but can't be started because the disk image is missing (it obviously wasn't moved).

Surprisingly, online migration works fine. One would expect it to be the other way around.

Proxmox: Virtual Environment 4.4-17/70a65945

Would appreciate any help, thanks in advance.

 

[wolfgang]

Hi,

this sounds like a miss configured storage conf.
Is the storage marked as "shared"?

 

[n3n5n1]

Yup, we have a "local" storage (/var/lib/vz) configured as "Shared", accessible from all nodes.

Quick note: to successfully migrate VM while keeping it online, we have to:
1) move its disk image to an NFS storage on a target node;
2) migrate VM with "Online" option checked;
3) move VM disk image back to "local" storage.

It seems like we should be good only with step #2. But it gives me an error:

Sep 19 16:55:22 starting migration of VM 114 to node 'xxx2' (xx.xx.xx.xx)
Sep 19 16:55:22 copying disk images
Sep 19 16:55:22 starting VM 114 on remote node 'xxx2'
Sep 19 16:55:23 volume 'local:114/vm-114-disk-1.qcow2' does not exist
Sep 19 16:55:23 ERROR: online migrate failure - command '/usr/bin/ssh -o 'BatchMode=yes' root@xx.xx.xx.xx qm start 114 —skiplock —migratedfrom nsrv3 —migration_type secure —stateuri unix —machine pc-i440fx-2.7' failed: exit code 255
Sep 19 16:55:23 aborting phase 2 - cleanup resources
Sep 19 16:55:23 migrate_cancel
Sep 19 16:55:24 ERROR: migration finished with problems (duration 00:00:03)
TASK ERROR: migration problems

 

[pabernethy]

Unless you actively share the "local" storage via NFS, Samba or other means, it's not shared. "Shared" does not mean "Share this to all nodes" but "I'm sharing this to all nodes"

 

[n3n5n1]

Okay, so I may be unclear on the exact procedure. Could you please point me towards a detailed description of online/offline migration mechanisms? The tools used, order of operations, etc.

 

[pabernethy]

You just have to mark the storage the disks reside on correctly. Actually shared storage as "Shared" and local storage as non-shared. Migration should automatically move the disk images on non-shared storage and keep the ones on shared storage.

 

[n3n5n1]

That cleared it up for me a bit. I've unchecked "Shared" option for my "local" storage and the offline migration started to work as expected.

Thanks a ton for your quick and helpful answers.