Severe system freeze with NFS on Proxmox 9 running kernel 6.14.8-2-pve when mounting NFS shares

K

Krailler

New Member
Aug 10, 2025
1
0
1
Hello everyone,

I have recently updated to Proxmox VE 9 with kernel 6.14.8-2-pve on an Intel N100 system and now, when mounting an NFSv4.2 share from a NAS server, performing write operations causes a massive increase in IO delay and eventually the entire system freezes, requiring a manual reboot.

Details:
  • The NFS server is working fine and the network is not saturated.
  • Other servers mounting the same NFS share don’t experience issues.
  • The mount uses default NFSv4.2 options with large rsize and wsize.
  • Tried different mount options (including NFSv3 and smaller sizes) with no success.
  • The problem only occurs on Proxmox 9 running kernel 6.14.8.
  • SMB mounts work fine without freezes.
  • Reinstalled nfs-common package without changes.
  • No ECC memory and no visible hardware issues.
Has anyone else experienced similar NFS freezes on Proxmox 9 with this kernel? Are there any known workarounds or fixes?

Thanks in advance.
 
We were able to reproduce a problem by mounting NFS-Shares with kernel "6.14.8-2-pve".
It is not possible to mount.

We tried to mount manually by using the following command:
Code: 
mount -t nfs -o vers=4.2 X.X.X.X:/pve/VMSTORE /mnt/pve/VMSTORE -vvvvvv
The following error occurs:

Code: 
mount.nfs: mount(2): Input/output error
mount.nfs: mount system call failed for /mnt/pve2/VMSTORE

After performing a rebooting with "6.8.12-13" everything works fine.
 
I am not only experiencing the above with NFS, but also with CIFS. Regular mounts work fine, but with any large data transfer, the IO delay comes and everything becomes unresponsive. The NFS/CIFS shares become disconnected and I have to reboot.

I'm running Proxmox 9 (kernel 6.14.8-2-pve) on a Minisforum MS-A2 with 96GB Ram. I never had issues like this on Proxmox 8. I tried looking for kernel 6.8.12-13, but it is unavailable on the Proxmox 9 free repositories. I haven't found any mention of these issues anywhere but here. I guess I either have to wait until the next kernel release, or re-install Proxmox 8 on my cluster.
 
I see the similar issue with my proxmox setup as mentioned by original poster when copying large amounts of data to nfs: high load average, lost connection to nfs vm, proxmox web UI does not get any data from VMs / unresponsive.

Not 100% sure, but it looks like I have found a way to 'unfreeze' the system in such case without the need to reboot the whole server - run `umount -f /mnt/pve/suspicious_nfs_share`, possibly several times. It won't unmount, because 'device is busy', but somehow it breaks deadlock and let the system continue to work. Maybe this is a coincidence, but this trick saved me a couple of times.
 
That does seem to unlock the mount. Strange that it's happening though. I didn't have an issue before upgrading to version 9, but when I reloaded one node with 8.4, it was doing the same thing. I can't figure it out.

Edit: That trick doesn't always work.
 
Last edited:
I was checking every configuration change that I did as my backups to NFS share was crashing. Looks like the update broke it huh :)
 
Mine is a TrueNAS Scale, which is running as a VM inside the same host. I didn't notice the freezing during usual NFS share usage, but my scheduled backups were failing as the NFS share was getting frozen after the update to Proxmox 9. I've had the same setup for a long time and didn't notice this issue previously. But SMB mount works fine, as stated by @Krailler
 
  • Like
Reactions: Gilberto Ferreira
harish2k_1 said:
Mine is a TrueNAS Scale, which is running as a VM inside the same host. I didn't notice the freezing during usual NFS share usage, but my scheduled backups were failing as the NFS share was getting frozen after the update to Proxmox 9. I've had the same setup for a long time and didn't notice this issue previously. But SMB mount works fine, as stated by @Krailler
Click to expand...
dmesg, journalctl -f show something?
 
I tried here with Proxmox 9.0.5 and kernel 6.14 and no issue! Using TrueNAS-SCALE-25.04.2.1.
The only think I saw was the in the web gui, I only could see the nfs share if I choose ver=3.
But after added, I had have edited the /etc/pve/storage.cfg and changed it to ver=4 and works fine, to both ISO and VM images.

Cheers.
 
You must log in or register to reply here.
Top Bottom