NFS hangs on system

[Jerry L. Sims]

Hello All,

Today our system ( rp7410 11i )hung when trying to access
NFS mounted system. The only error mesages I
could locate were the following :

dmesg & syslog reported the following:
Apr 1 09:23:44 it-hp56 vmunix: NFS server psh-cel-06 not responding still trying

I attempted to execute :
/sbin/init.d/nfs.client stop
which just hung

I attempted to unmount & mount the nfs filesystem, but received a response it was busy. The "fuser -u" command indicated nothing
had it busy. I attempted to /etc/shutdown -r now, and the system hung. The /etc/rc.log.old at the following :
Stop NFS client subsystem
Output from "/sbin/rc1.d/K570nfs.client stop":

I had to power reset the system. Please advise.

Also, I checked (3) other hp severs who had the
same nfs server mount point working fine.

[Todd McDaniel_1 · Accepted solution]

When you have NFS mount problems as I do occasionally, You can do a few things short of rebooting.

1) stop automount and then remove entry from /etc/fstab and rm /etc/mnttab and then umount it.

2) if you do want to reboot the box, simply do a "reboot" not a shutdown, it will bypass the rc* scripts.

[Martin Johnson]

We experienced the same thing today on an AIX server that had NFS mount HPUX file systems. IBM recommended reboot.

Normal shutdown tries to umount file systems, which will hang. You need to do the reboot command, which is a little more graceful than a reset system.

HTH
Marty

[Hoefnix]

Using the reboot can get you in more trouble if the system is for example running a database.

Better is to rename some rc scripst by tempory rename for example the NFS kill scripts in the /sbin/rc.d directories. Rename some involved K???scripts to _K???scripts and they will not be executed during shutdown.

If you want to disable some scripts during startup do the same with the S???scripts

Regards,
Peter

[Todd McDaniel_1]

You can go to init 2 first and then reboot most DB rc scripts will be executed on /sbin/rc2.d/KXXX...

[Hoefnix]

Todd, you are right, that will mostly do the trck as well. It was just an extra thought to remind Jerry to check this before reboot.

Regards,
Peter

[SS_6]

Your problem looks like only on client side side or nfs client side, still check nfs server side.
1.ping psh-cel-06 and see nfs client is able to talk.
2.Check your NFS server psh-cel-06 is sucsessfully exporting the file systems to this client.
#exportfs
If problem is on the client side, you can soft kill automounter demon and start again or
/sbin/init.d/nfs.client stop
/sbin/init.d/nfs.client start
(You may recycle other nfs.core, nfs.server on client side)
Otherwise I would recommend to do a reboot.
Always stop automounter to make any changes to configuration ofnfs file systems.

[Jerry L. Sims]

what is the meaning of "soft kill automounter" ? Would that be to :
1) # ps -aef | grep auto
root 735 0 0 10:13:45 console 0:00 autofs_proc
root 732 1 0 10:13:45 ? 0:05 /usr/lib/netsvc/fs/autofs/automountd

2) kill -9 732

3) ]# /usr/lib/netsvc/fs/autofs/automountd

?

[SS_6]

Soft Kill,I mean to say without -9 option. Even if you look at the nfs.client script. it is stoping and starting automountd. I have seen in 80% cases if nfs hung only we have to reboot.