SUMMARY: fake out NFS clients to prevent Stale File Handles

From: Lucas, Jeffrey (
Date: Thu Feb 03 2000 - 15:10:44 CST

Thank you all responders!

Original Question:
I've searched the archive and wasn't able to find an answer to the following
(although there were several msgs regarding troubleshooting stale file

I have an NFS share with several NFS client mounts. I want to unmount the
filesystem being shared in order to back it up, resize it, restore it,
reshare it while not 1) having to unmount all the clients, and 2) causing
the clients to end up with stale file handles.

Does anyone know if this is possible and if so, how to do it?

I realize this can be dangerous but the NFS share is not critical for
clients. I just don't want to unmount all those clients or end up with stale
file handles.

Thanks in advance. Will summarize.

Summary of Responses (Unfortunately, no quick and easy way.
1) As I am dealing with a ufs filesystem on a disk partition (vs. an
encapsulated disk in Volume Manager / Disk Suite), I will be unable resize

2) The re-partitioning/labeling of the disk as well as the newfs will result
in *new inode numbers*, which NFS is dependent upon. Unless I could preserve
inode numbers, it's likely that the clients would not be happy (i.e. stale
file handles).

3) The safest/cleanest way seems to be to the most logical... unmount the
shared filesystem from each client, unshare the filesystem from the NFS
server, backup the data, repartition/relabel the disk, newfs, restore data
and reshare filesystem.

Thanks again!

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:14:02 CDT