[PVE-User] PVE 5.4: cannot move disk image to Ceph

[Uwe Sauter]

Am 06.09.19 um 12:32 schrieb Alwin Antreich:
> On Fri, Sep 06, 2019 at 11:44:10AM +0200, Uwe Sauter wrote:
>> root at px-bravo-cluster:~# rbd -p vdisks create vm-112-disk-0 --size 1G
>> rbd: create error: (17) File exists
>> 2019-09-06 11:35:20.943998 7faf704660c0 -1 librbd: rbd image vm-112-disk-0 already exists
>>
>> root at px-bravo-cluster:~# rbd -p vdisks create test --size 1G
>>
>> root at px-bravo-cluster:~# rbd -p vdisks ls
>> test
>> vm-106-disk-0
>> vm-113-disk-0
>> vm-113-disk-1
>> vm-113-disk-2
>> vm-118-disk-0
>> vm-119-disk-0
>> vm-120-disk-0
>> vm-125-disk-0
>> vm-125-disk-1
>>
>> root at px-bravo-cluster:~# rbd -p vdisks rm test
>> Removing image: 100% complete...done.
>>
>> root at px-bravo-cluster:~# rbd -p vdisks rm vm-112-disk-0
>> 2019-09-06 11:36:07.570749 7eff7cff9700 -1 librbd::image::OpenRequest: failed to retreive immutable metadata: (2) No such file or
>> directory
>> Removing image: 0% complete...failed.
>> rbd: delete error: (2) No such file or directory
> Seems Ceph has the vm-112-disk-0 still stored somewhere. At least this
> error message should be visible in the ceph logs. Hopefully there is
> more info about it.
> 
> Does the 'rbd showmapped' show the image being mapped? Are you running
> filestore or bluestore OSDs?
> 
> If you migrate the VM to a different host, does it occure there too?
> 

Neither Ceph had vm-112-disk-0. I solved it by disabling the second Ceph instance. PVE was then able to move the disk.

Yes, it occured on all hosts.


Thanks,

	Uwe



> --
> Cheers,
> Alwin
>