[SOLVED] qemu/kvm: 'gluster' is deprecated ?

Gilberto Ferreira

Renowned Member
Hi

Is the gluster not support properly anymore?
I set two PVE box with Proxmox VE 8.4 and GlusterFS 10, which is shipped with Debian Bookworm.
I have had used gluster for many years now, with no issues at all.
You guys will, eventually kick off GlusterFS for good?

More warning messages:
generating cloud-init ISO
qemu-img: warning: 'gluster' is deprecated
[2025-04-10 16:30:25.881891 +0000] I [io-stats.c:3784:ios_sample_buf_size_configure] 0-VMS: Configure ios_sample_buf size is 1024 because ios_sample_interval is 0
[2025-04-10 16:30:26.531746 +0000] E [MSGID: 108006] [afr-common.c:6103:__afr_handle_child_down_event] 0-VMS-replicate-0: All subvolumes are down. Going offline until at least one of them comes back up.
[2025-04-10 16:30:35.885957 +0000] I [io-stats.c:4190:fini] 0-VMS: io-stats translator unloaded
kvm: -drive file=gluster://gluster1/VMS/images/100/vm-100-cloudinit.qcow2,if=none,id=drive-ide2,media=cdrom,format=qcow2,aio=io_uring: warning: 'gluster' is deprecated
TASK OK
 
Last edited:
[ SOLVED ]

I just change from the GlusterFS plugin to directory pulgin and the warning message is gone.
I usually to a gluster volume and define it inside /etc/fstab file, like this:

serverA:
gluster1:VMS /vms glusterfs defaults,_netdev,x-systemd.automount,backupvolfile-server=gluster2 0 0

serverB:
gluster2:VMS /vms glusterfs defaults,_netdev,x-systemd.automount,backupvolfile-server=gluster1 0 0

And then mounted the /vms directory with is mapped to gluster volume.
After that, I just define the storage type as directory, mapped to that /vms directory, just like that:

dir: VMS
path /vms
content images
preallocation off
prune-backups keep-all=1
shared 1

Cheers
 
  • Like
Reactions: waltar
Hi,
yes, unfortunately GlusterFS is unmaintained and it will be dropped at some point from upstream QEMU. It will not be dropped during the Proxmox VE 8.x life cycle however.
 
  • Like
Reactions: Gilberto Ferreira
Hi,
yes, unfortunately GlusterFS is unmaintained and it will be dropped at some point from upstream QEMU. It will not be dropped during the Proxmox VE 8.x life cycle however.
Well... It's ok. Sad, but ok.
Usually I create a gluster volumes and mount both nodes.
Then, I share this as type directory storage. Never use the glusterfs plugin.
So, for qemu/kvm and proxmox standpoint, is always a directory, right?
Like I do here
Anyway, I don't think gluster is dead! Just get slow for a while. I saw some active in github and even some comments about version 12.
Will see in the future.
 
Last edited:
Well... It's ok. Sad, but ok.
Usually I create a gluster volumes and mount both nodes.
Then, I share this as type directory storage. Never use the glusterfs plugin.
So, for qemu/kvm and proxmox standpoint, is always a directory, right?
Like I do here
Anyway, I don't think gluster is dead! Just get slow for a while. I saw some active in github and even some comments about version 12.
Will see in the future.
If you don't use the GlusterFS storage plugin, but a shared directory storage, then the image will also not be attached via the gluster protocol driver to QEMU. So that will still work in future Proxmox VE 9.x releases :)
 
  • Like
Reactions: waltar