Stale file handle (ESTALE) is NFS's way of saying "the thing this handle pointed to is gone." It shows up when a server is re-exported differently, loses its export, or is replaced — and clients are left holding references the server no longer honors, so reads hang or error out.
Check the server's exports first
exportfs -v # what the server is actually exporting # client side: cat /proc/mounts | grep nfs # server log: "rpc.mountd: refused mount request: no export entry"
Fix the export table and re-export
# /etc/exports /export/data 10.0.0.0/24(rw,sync,no_subtree_check) exportfs -ra # re-read /etc/exports and apply
Once the server exports correctly, clients still holding stale handles may need a lazy remount (umount -l then mount -a) — historically a client reboot was the only cure. Prefer hard mounts for data you can't lose: they hang and recover rather than returning errors mid-write.
Practise it
Broken NFS exports are one of the simulator's storage/file-service incidents. Work them in a full incident response shift.