NFS problem

[Mehdi_1]

Hi

Some of the discs on our nfs server (HPUX Server) is not seen by some of our HPUX client ( This was working ok this morning ).

As a result I tried to stop and start the nfs.client on one these clients machine. But I got the error shown below.

Then I rebooted the same client machine with the hope fixing the problem, but it didn't fix anything at all, and the same problem is still there.

Considering nothing has changed since this morning, it is very strange to see problem.

Any help is appreciated.

Thanks

__Mehdi

killing rpc.lockd
killing rpc.statd
umountall: umount : has failed.
umountall: diagnostics from umount
nfs umount: nfs_unmount: /home/workexp: is busy
umountall: umount : has failed.
umountall: diagnostics from umount
nfs umount: nfs_unmount: /net/usr: is busy
umountall: umount : has failed.
umountall: diagnostics from umount
umount: cannot unmount /home : Device busy
umountall: umount : has failed.
umountall: diagnostics from umount
umount: cannot unmount /net : Device busy
killing biod
killing automountd

[Pete Randall]

I would start investigating the NFS server. Are there any problems apparent there? Are all of the exported file systems ready and available. Do you have any disk problems (check dmesg output and /var/adm/syslog/syslog.log)?


Pete

[Zeev Schultz]

Looks like some stale mounts and / or locks
problem.But all NFS client side daemons are
dead - lockd,statd,biods...Ok,this is natural
to nfs client to claim back his mounts after
hang/reboot...
What mount -v shows?What is on server side about this disk in ioscan -fnCdisk and syslog?

Zeev

[Massimo Bianchi]

Hi,
check and clean the /etc/rmntab, you can find funny entried in it after a dirty shutdown...

Massimo

[Mehdi_1]

There is no problem with the disks. It can bee seen by other HPUX clients. The disk is functioning ok and work perfect.
From about 10 clients 6 of them can see the disk and everything works ok, only 4 clients can see the disk!!!

Thanks

__Mehdi

[Zeev Schultz]

Run showmount -a on server,check if it sees these clients.Start nfs.client on clients,try to run umount on filesystems.Do cat /etc/mnttab
before that to see what appears there.
I'd also clean /var/statmon files (stop lockd,statd before)
rm -rf /var/statmon/sm.back , state.back
mv /var/statmon/sm sm.back , state to state.back.

Zeev

[Steven E. Protter]

If you want to do a complete restart of NFS on the server you need to do this:

/sbin/init.d/nfs.client start
/sbin/init.d/nfs.server start
/sbin/init.d/nfs.core start

/sbin/init.d/nfs.client start
/sbin/init.d/nfs.server start
/sbin/init.d/nfs.core start

SEP

[Mehdi_1]

Hi

Thanks for all the replies.

The problem hasn't been solved yet.

I 've restarted the NFS server and I also cleaned the "/var/statmon". The command showmount -a shows that the directory is been exported to the client machines, But some of the client machines can't see it.

When I rebooted one of the client machine, in boot time I got the error:

Start Software Distributor agent daemon ... Failed

Configuring all unconfigured software filesets ... Failed

I am not sure that has got anything to do with my current problem.

In file "/etc/rc.log" shows:

ERROR: Cannot contact the 'swagentd' daemon because the local
hostname 'computer name' cannot be resolved.

Thanks again

__Mehdi

[Dagmar Boelen]

Stupid questions maybe. Can all the clients ping the NFS-server?

[Pete Randall]

The swagentd problem sounds like a hostname resolution issue. Can you resolve your hostname successfully, both by name and by ip, using nslookup? Are your /etc/nsswitch.conf and /etc/resolv.conf files set up correctly? Maybe you should run /sbin/set_parms hostname to make sure all the possible locations for hostname are set to the same value.

This could well be related to your NFS issues!


Pete

[Mehdi_1]

Yes, All the clients can ping sever and also server can ping all the cilent machines.

Also the nslookup works ok and can resolve the IP or the host names in both server and client sides.




__Mehdi