CephFS mount breaks with network connectivity interruption, does not fix when network resumes

[BloodyIron]

Tried to GoogleFu and didn't seem to find anyone else having this problem.

Relevant PVE cluster is like 11 nodes, 3x of which run a Ceph cluster serving RBD and FS.

All nodes in the cluster use the RBD storage and FS storage. VMdisks are allowed for RBD but not FS, FS is primarily ISOs in this case so not a big deal but still hoping to find a "solution" here.

When one of the nodes not running the Ceph cluster loses network connectivity for $reasons, then regains connectivity, the RBD network storage comes back up just fine, but the CephFS mount breaks and never fixes itself. A reboot fixes it of course, but that's silly. If I tell the CephFS mount on the PVE Node to unmount that corrects it (assuming I don't have any ISOs mounted by VMs running on that node).

But I'm hoping I can find a way to make it so CephFS fixes itself when this happens... without writing a script or something like that.

Anyone have any advice on what I could do about this, while drinking the PVE kool-aid? I'd prefer to use a PVE-doc'd method if possible. Either way my ears are open Thanks!

 

[jeinwag]

I think the recover_session=clean mount option may be what you're looking for.

 

[BloodyIron]

I think the recover_session=clean mount option may be what you're looking for.

I'm not entirely sure what syntax/structure to use to modify the /etc/pve/storage.cfg to apply this, and the webGUI does not offer a nice field for adding mount options. How should I modify said config file to make that work?