Need Help to Fix Stale NFS File Handle

[Mark Phillips]

On Fri, Feb 25, 2011 at 4:57 PM, Alex Dean <alex at crackpot.org> wrote:

>
> On Feb 25, 2011, at 4:35 PM, Mark Phillips wrote:
>
> >
> >
> > On Fri, Feb 25, 2011 at 3:28 PM, Alex Dean <alex at crackpot.org> wrote:
> >
> > On Feb 25, 2011, at 4:16 PM, Mark Phillips wrote:
> >
> > >
> > >
> > > On Fri, Feb 25, 2011 at 2:30 PM, Alex Dean <alex at crackpot.org> wrote:
> > >
> > > On Feb 25, 2011, at 3:13 PM, Mark Phillips wrote:
> > > > # mount | grep nfs
> > > > return anything?
> > > > hammerhead:/home/mark# mount | grep nfs
> > > > rpc_pipefs on /var/lib/nfs/rpc_pipefs type rpc_pipefs (rw)
> > > > hammerhead:/home/mark#
> > > >
> > > > I have no idea who rpc_pipefs on /var/lib.... is and why it is there!
> Any ideas before I blow it away?
> > >
> > > That's NFS-server related.  Used by the id mapper, I think.  On RedHat
> at least, I think that's started/stopped by /etc/init.d/nfslock. (Though my
> memory is a little sketchy on that point.)  The pipefs could possibly be
> mounted even if nothing's using it, also.  Got any NFS-related services
> running?
> > >
> > > Afaik, an NFS server shouldn't get stale file handles.  That's a
> client's problem.
> > >
> > > RedHat/Fedora:
> > >  $ rpm -qa | grep nfs
> > > Debian/Ubuntu:
> > >  $ dpkg --list | grep nfs
> > > Both:
> > >  $ ls /etc/init.d | grep nfs
> > >
> > > hammerhead:/home/mark# dpkg --list | grep nfs
> > > rc  libnfsidmap1                         0.8-1
>        An nfs idmapping library
> > > ii  libnfsidmap2                         0.20-1
>       An nfs idmapping library
> > > ii  nfs-common                           1:1.1.2-6lenny2
>        NFS support files common to client and serve
> > >
> > > hammerhead:/home/mark# ls /etc/init.d | grep nfs
> > > mountkernfs.sh
> > > mountnfs-bootclean.sh
> > > mountnfs.sh
> > > nfs-common
> > > umountnfs.sh
> > > hammerhead:/home/mark#
> > >
> > > Wow...I had no idea nfs was running. I have no idea how it is being
> used or by what process. Do I stop it with /etc/init.d/umnountnfs?
> > >
> > > Mark
> >
> > unmountnfs.sh is part of initscripts.  I don't think you're expected to
> call it directly, so I'd be inclined to say leave it alone.  I think you can
> try removing all those packages if you really don't need them.  To stop
> nfs-common, you'd use '/etc/init.d/nfs-common stop'.
> >
> > alex at artichoke:~$ cat /etc/issue
> >  Ubuntu 9.10 \n \l
> > alex at artichoke:~$ dpkg -S /etc/init.d/umountnfs.sh
> >  initscripts: /etc/init.d/umountnfs.sh
> >
> > You can use "apt-cache showpkg nfs-common" to figure out if you've got
> any packages which depend on nfs-common.  (Same for libnfsidmap1 and 2.)
>  Look in the 'Reverse Depends' section.  Or just "apt-get remove ..." and
> see what warnings you get.
> >
> > I imagine the uninstallation scripts should umount the pipefs filesystem,
> but that may not be true.  Either way, I don't think that's the cause of
> your 'stale file handles' issue, but I could certainly be wrong on that
> point.
> >
> > What if I just delete the files that have stale NFS file handles and
> re-install them?
> >
> > Mark
>
> Then you have no idea if the problem will recur or not.  (If it happened
> once, it probably will again.)  Did you ever check if you have automount
> running?
>
> Take a look for automount's config files and see if anything seems
> familiar.
> $ ls /etc/auto*
>

hammerhead:/home/mark# ls /etc/auto*
ls: cannot access /etc/auto*: No such file or directory


> $ cat /etc/auto.master
>
Don't have one of those beasts...It doesn't look as if I have automount
running.

Could the stale file handles be caused by the disk controller card failing?
And then installing a new card?

Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.PLUG.phoenix.az.us/pipermail/plug-discuss/attachments/20110225/5664c02a/attachment.html>