NFS subdirectory mounts cause stale file handle errors

If you mount a subdirectory of an NFS export rather than the root of the share, you’ll get stale file handle errors when the NFS server restarts or the share is unmounted and remounted on the server side.

Don’t do this:

192.168.1.10:/volume1/data/appdata  /mnt/appdata  nfs  defaults  0 0

Do this instead — mount the root share and use a subdirectory locally:

192.168.1.10:/volume1/data  /mnt/nas  nfs  defaults  0 0

Then symlink or use /mnt/nas/appdata directly. The stale handle errors occur because the server’s file handle for a subdirectory becomes invalid when the parent export changes state, even if the subdirectory itself is untouched.

fstab options that matter

• _netdev: tells the OS this mount requires the network. Without it, the system tries to mount during early boot before networking is up and hangs.
• nofail: allows the system to boot even if the mount fails. Essential for anything you don’t want causing a boot hang.
• x-systemd.automount: lazy mount — only actually mounts when something accesses the path. Useful for rarely-used shares.

NFS failover with autofs

Autofs supports replicated server syntax for automatic failover between two NFS servers:

/mnt/nas  -fstype=nfs,soft,timeo=5  server1:/share  server2:/share

Autofs tries the first server and falls back to the second if it doesn’t respond within timeo. Failover takes 5–30 seconds — fast enough for most homelab use cases without the complexity of a floating VIP.