[SOLVED] Cannot migrate VM "Failed to sync data - unable to parse directory volume name"

[Klug]

Hello.

I have this issue with two VM (maybe more).

They were migrated from a vSphere setup.
I first moved them (while in vSphere) them to a shared NFS path (shared by vSphere and Proxmox).
Then stopped them in vSphere, created the "same" VM in Proxmox (using vmdk format for hard drive).
Then replaced the Proxmox-created vmdk by the vSphere vmdk and copied the -flat.vmdk file to the same directory.
Eventually started the VM in Proxmox and change the storage to a Ceph volume.

I've done this for dozens of VM and it worked well.

Hower, at least two can not be migrated from one node to the other in the Proxmox cluster.

Here's the error when trying to migrate one:

2021-03-08 18:07:11 starting migration of VM 202 to node 'pve02' (192.168.88.152)
2021-03-08 18:07:12 ERROR: Failed to sync data - unable to parse directory volume name '202/ZCS - OSS-flat.vmdk'
2021-03-08 18:07:12 aborting phase 1 - cleanup resources
2021-03-08 18:07:12 ERROR: migration aborted (duration 00:00:01): Failed to sync data - unable to parse directory volume name '202/ZCS - OSS-flat.vmdk'
TASK ERROR: migration aborted

The error is about the old VMDK file on the shared NFS shared storage.
BTW, the file is still on the this volume:

# cd /path/volume/images/202
# ls -l
total 176603050
drwxr-----   2 root  wheel  uarch            4 Feb 13 10:47 ./
drwxr-xr-x  13 root  wheel  uarch           13 Feb 13 18:56 ../
-rw-r-----   1 root  wheel  uarch     33685504 Feb 12 23:04 vm-202-disk-0.vmdk.orig
-rw-------   1 root  wheel  uarch 274877906944 Feb 13 10:47 ZCS - OSS-flat.vmdk

The VM config file seems OK, no more vmdk file in it but "SSD" as storage (that's the Ceph volume):

# qm config 202
agent: 1
boot: order=scsi0;ide2;net0
cores: 4
ide2: none,media=cdrom
memory: 8192
name: ZCS-zimbra-oss
net0: virtio=22:82:5F:C8:20:AB,bridge=vmbr0,tag=101
numa: 0
ostype: l26
scsi0: SSD:vm-202-disk-0,discard=on,size=256G
scsihw: virtio-scsi-pci
smbios1: uuid=90434022-0b3d-464c-b0ef-231fb525b287
sockets: 1
vmgenid: 5c29ac94-a16a-47f7-96b3-15afc35076f7

Any idea what the problem could be?

 

[Klug]

Funny one.

I renamed the 202 folder on the NFS storage to 020. Just to try if it would do anything.
It fixed the issue, the VM can not be migrated.

Looks like we have to manually remove any left folder on old storage and migration task does not rely on the "qm config" on the VM but parses folders...

Should I file a bug somewhere?