Gluster 3.7.12 and qcow disks

[Ulrar]

Hi,

Gluster 3.7.12 came out yesterday, and since 3.8 isn't compatible with proxmox yet I installed 3.7.12.
I used XFS for the bricks like I always did, and I was able to add it in proxmox just fine (where 3.8 is a bit wonky to add in proxmox).

Now when creating a new VM I can't actually use qcow, if I do the creation will timeout with that error in the logs :

Jun 29 13:26:25 s2.name pvedaemon[2803]: create failed - unable to create image: got lock timeout - aborting command

Jun 29 13:26:52 s2.name qemu-img[2811]: [2016-06-29 11:26:52.485296] C [rpc-clnt-ping.c:165:rpc_clnt_ping_timer_expired] 0-gluster-client-3: server 172.16.0.2:49153 has not responded in the last 42 seconds, disconnecting.

Jun 29 13:26:52 s2.name qemu-img[2811]: [2016-06-29 11:26:52.485407] C [rpc-clnt-ping.c:165:rpc_clnt_ping_timer_expired] 0-gluster-client-4: server 172.16.0.3:49153 has not responded in the last 42 seconds, disconnecting.

Jun 29 13:26:52 s2.name qemu-img[2811]: [2016-06-29 11:26:52.485443] C [rpc-clnt-ping.c:165:rpc_clnt_ping_timer_expired] 0-gluster-client-5: server 172.16.0.50:49153 has not responded in the last 42 seconds, disconnecting.

But I can use raw disks just fine. I noticed when trying to use qcow that the file does get created, and is left on the gluster after the timeout.
I googled a bit and found that the same thing was happening in 2014 for people using XFS for their bricks so I changed mine to use ext4, same problem.

Ofcourse, the volume works just fine when using it by hand, only proxmox (or qemu ?) seems to have a problem with it.


EDIT : Spoke a bit too soon, raw does seem to work and can boot, but when trying to actually use the disk in the VM (mkfs.ext4) it just shuts down without any error.
Everything works fine using NFS though, I'll use that until it's fixed. That way I can even use qcow