[SOLVED] vDisk "move" command skips over "thin provisioning" on Ceph_Pools ??

[Q-wulf]

So I just noticed something I never noticed before and I am not sure wether it is a bug or not.

I have 2 ceph pools.
Pool Ceph_A and Ceph_B for simplicities sake. They are both Erasure Coded pools with a cache pool infront of it (afaik mandatory for EC-pools)

Now because i had issues deleting a different vDisk (Known issue: "image still has watchers") i decided to move all Disks off Ceph_A and onto Ceph_B that i wanted to keep.

On Ceph_A i have setup a vDisk using SCSI (Discard=on) using virtio (scsi controller type). size is 4096 GB. 66 GB are utilized. i used the gui to move said vDisk to Ceph_B pool and now i have a 4096GB utilisation on said pool. The thin provisioning seems to have gone. Or in other words, my 66GB image turned into a 4096 GB image via the "move disk" command on GUI.

Is that working as expected, or is this a bug ?

 

[spirit]

They are missing qemu feature for rbd block driver, for drive mirroring.
(It's copy all block, including zeros)

As you use discard=on, you can do a trim inside your guest vm to free space.
(you can use "fstrim" command fo example.

 

[spirit]

For the reference, I had send a mail to qemu mailing about this in 2014
http://qemu.11.n7.nabble.com/qemu-drive-mirror-to-rbd-storage-no-sparse-rbd-image-td293642.html

"

Lack of bdrv_co_write_zeroes is why detect-zeroes does not work.

Lack of bdrv_get_block_status is why sparse->sparse does not work
without detect-zeroes.

"

 

[Q-wulf]

so, this is "working as intended" then.


Its a usecase, that never comes up in our production use.
It just came up on my private nodes, so i thought i mention it.

Thanks @spirit