Proxmox VM freezes during NFS backup? It’s the snapshot hanging on I/O. The VM pauses while qemu waits for the NFS server to flush data, and if the server is slow or the mount is misconfigured, that wait can be endless. No errors in the logs, just a frozen VM.
Check the VM process state
First, figure out if the VM is actually frozen or just slow. Run qm status <vmid> on the Proxmox host. If it says running, the VM is alive but stuck. Then look at the process with ps aux | grep <vmid>. You’ll see the qemu process. If it’s in D state (uninterruptible sleep), it’s waiting on I/O. That’s your smoking gun.
Next, check /proc/<pid>/stack for that qemu process. You’ll likely see something like nfs_wait_on_request or rpc_wait_bit_killable. That confirms the VM is stuck waiting for NFS to respond.
Why NFS snapshots hang
When Proxmox takes a snapshot for backup, it freezes the VM’s filesystem and flushes all pending writes to the storage. With NFS, that flush goes over the network. If the NFS server is slow, or the mount is using sync exports, the flush can take a long time. Worse, if the NFS server stops responding, the VM hangs indefinitely.
Common culprits:
- NFS server overloaded or slow disks
- Network issues between Proxmox and NFS server
- NFS mount options like
hard(default) that retry forever - Using NFS over a high-latency link
Fix the freeze
First, check your NFS mount options on the Proxmox host. Look at /etc/pve/storage.cfg for your NFS storage. If you see options hard, that’s the problem. Change it to soft,timeo=50,retrans=2 to make the mount fail after a few seconds instead of hanging forever. You can edit the file directly or use pvesm set <storage-id> --options soft,timeo=50,retrans=2.
If you can’t change mount options, try reducing the backup load. Use ionice to lower the backup process priority, or schedule backups during off-peak hours. Also, make sure the NFS server has enough resources and the network is stable.
Another thing: if you’re using snapshot mode in vzdump, try suspend mode instead. That pauses the VM completely before backup, which avoids the flush hang but causes downtime. Not ideal, but it’s a workaround.
My take
This is a stupid problem because it’s silent. The VM just freezes and nothing in the Proxmox logs tells you why. You have to dig into the process state. Once you see D state and NFS in the stack, it’s obvious. I’d fix the mount options first—soft mounts are less reliable but at least they fail instead of hanging. For a homelab, that’s fine.
If you’re dealing with other Proxmox storage issues, check out proxmox backup skips bind-mounted NAS datasets in unprivileged LXC containers.
