Are there ways of getting rid of a stale NFS handle without rebooting?
All ideas welcome.
Has anybody had problems using the pfsmount utitlity and NFS errors
Just open for general discussion.
>> Are there ways of getting rid of a stale NFS handle without
"Michael> If a filesystem has gone away, rebooting is the only
"Michael> thing I've found that works. If the filesystem comes
"Michael> back, the mount will come back.
But a filesystem that causes 'stale filehandle errors' will
not work again -- I believe NFS filehandles are not reused.
In my experience the only way to clean up mountpoints that
cause 'stale NFS filehandle' errors is to reboot the client system.
BTW - what is pfsmount?
Quentin Fennessy AMD, Austin Texas
Technical Consultant | I speak for me, | 19055 Pruneridge Ave.
Development Alliances Lab| *not* HP | MS 46TU2
IASD / Enterprise Systems Group +--------------+---- Cupertino, CA 95014
While the following may not apply in the case of the original poster,
what you can sometimes do instead of rebooting is to simply stop/restart
the NFS deamons and any automounting on the client machines involved - in
this case the machine with the remote mounted disk and any client machines
that remote mount it. The scripts nfs.server and nfs.client in
sbin/init.d will do this (HPUX10 and HPUX11). This is what you are doing
when you reboot. It doesn't always work, particularly if some process
has opened a file on a stale NFS mount but you can track these proceses
down using the fuser command and then kill them before running these
Of course, rebooting everything is always available as a last resort :-)
> Actually, I believe this was fixed by in the NFS patches in the last
> year or so.
> Technical Consultant | I speak for me, | 19055 Pruneridge Ave.
> Development Alliances Lab| *not* HP | MS 46TU2
> IASD / Enterprise Systems Group +--------------+---- Cupertino, CA 95014
> > >If a filesystem has gone away, rebooting is the only thing I've
> > >found that works. If the filesystem comes back, the mount will
> > >come back.
> > Actually, I believe this was fixed by in the NFS patches in the last
> > year or so.
> > Rich
I just got thru goint thru each server and client, and killed the
'nfsd' and 'biod' daemons, then restarted the 'nfsd' only on the NFS
servers (it was running on some clients), andf the 'biod' on all
After doing all this, I just noticed the dreaded "stale handle" error
message on one machine. Also just then, I happened to check ping
betwee that machine and the NFS server- and got NADA ! It was a few
minutes before ping would work, and curiously, NFS mounts became OK.
So I wonder if the fault is in the netwreork somewhere, vice the
QUESTION: Should I run BIOD on the NFS servers as well, even if they
don't mount any NFS drives?
QUESTION: Where else should I look for the cause of continuing
'stales', assuming I can rule out the servers, clients ?
Thanks in advance for any help
--== Sent via Deja.com http://www.deja.com/ ==--
---Share what you know. Learn what you don't.---
After a power outage (actually 2) I am in trying to set the NFS mounts
back up. Everything is going smoothly until I get to the HPUX machine.
I try to mount /accts and I get:
mount: /accts: Stale NFS file handle
So I check out the faq and it says to unmount it hen remount so I try to
unmount it and get :
umount: /accts: Stale NFS file handle
I try to remove the directory but I m told that I can't. Any help would
be greately appreciated
Concentus Technology Corporation
5115 Parkcenter Ave, Suite 150
Dublin, Ohio 43017
Phone (614) 799-4317