trully umount stale filesystems cifs || nfs

M

migs

New Member
May 8, 2024
2
0
1
I encontered this problem several times with several remote filesystems and i really need help.
I had to change the password In a CIFS server which i only use for backups, the problem is the filesystem went away and now i am getting flooded in the logs and dmesg with:

[133862.953522] CIFS: Status code returned 0xc000006d STATUS_LOGON_FAILURE
[133862.953530] CIFS: VFS: \\10.10.10.10 Send error in SessSetup = -13
[133863.602243] CIFS: VFS: No writable handle in writepages rc=-9
[133863.603042] CIFS: VFS: No writable handle in writepages rc=-9
Click to expand...

even tough i re-created the mount point now i am getting flooded

How do i fix this problem without a full server restart?
I had encontered these issues in other network filesystems like nfs, and specially s3 fuse...
when there is not something right, i am forced to restart the whole server which is really awfull

i did tried to umount, forcefully umount the cifs storage and there is nothing that seems to help getting rid of the problem unless i restart the machine...
 
Last edited:
Kingneutron said:
https://stackoverflow.com/questions/74626/how-do-you-force-a-cifs-connection-to-unmount

I ran into the same issue a while ago, ' umount -l ' (that's an L) should do the trick
Click to expand...
that's what caused the biggest issue,
i did the lazy umount, the flood didn't stop and i had no way to get the mountpoint back again, even thought i recreated the mouting point.
i also tried to kill kill the processes and there was no use...

getting the proxmox server unstable and with the dmesg getting flooded because a network storage drive went away, is no fun...
and as i mentioned this happened to me in different ocasions when i was setting up storage networks...
because something trivial like changing a password, a temporary disconnect, or just initial testing to setup a new storage...
i am starting to get to the conclusion to not touch anything network storage related in the proxmox server..
 
I think this is why most ppl recommend mounting the shared storage in a container and sharing it out from there, that way a flood doesn't impact you at the host level

But yah, if it works don't mess with it
 
I had stale mounts earlier today tinkering with Ceph .. Doing a rolling reboot of all cluster nodes got things ok again. I know you said you didn't want to restart but sometimes that's just the cleanest way to handle it
 
You must log in or register to reply here.
Top Bottom
Back