Disk migration from to Ceph storage occupies the maximum capacity configure for the disk at once

[raydel92]

I have one issue...

When I migrate a VM's disk from one Ceph storage pool size 2 to another Ceph storage pool size 3 it occupies the maximum capacity configure for the disk in the target Ceph storage, despite the real use of that disk is only a few gigabytes. The attached images are very illustrative.

Anyone has encountered an issue like this?

Thanks,

Raydel

 

[chotaire]

Did you try to trim afterwards? E.g.

fstrim -a -v

 

[raydel92]

Did you try to trim afterwards? E.g.

fstrim -a -v

Should I do it in the host or in the guest?

 

[hungerfisch]

guest

 

[raydel92]

Did you try to trim afterwards? E.g.

fstrim -a -v

Didn't work. In fact I did not get any Verbose. I have my VMs running on distributed storage Ceph, perhaps that that command does not work with this.

 

[chotaire]

Is the drive for this VM configured as virtio-scsi with discard enabled?

 

[raydel92]

Is the drive for this VM configured as virtio-scsi with discard enabled?

I enable discard two days ago, the SCSI Controller is VirtIO SCSI and the Bus/Device of the disk is VirtIO Block. I think this last option cannot be changed, at least in the GUI.

 

[raydel92]

I did some additional tests with other VM’s virtual disks migrating them with the VM turned off and on. The results were that when I migrate the disk with the VM turned on it consumes the hole disk capacity in the target storage, instead when I did it with the VM turned off it just consumes the occupied capacity of the virtual disk in the target storage. In the future, I won’t be moving VMs from storage but there most be a solution to my current issue instead of turning the VM off, migrate the disk, and then turn it on again. In the image attached you can see the results of the test, left arrow (disk migration live) right arrow (disk migration with VM turned off).

 

[raydel92]

Any ideas? I need to reduce the fake 100GB occupied to the real 4GB of the VM