nfsktcpd process getting hanged

[Subentu]

Guys,

I have a Rx7620 server running HPUX 11.23.

Suddenly the process "nfsktcpd" shows a very high CPU usage in top output (around 100% and sometimes about 120%).

This level stays for about half an hour before coming back to the normal level.

When the CPU usage is high, the system gets really slow and the bdf output hangs for a couple of seconds at the nfs mounted file systems.

Can anyone please help?

Thanks

[Analyst]

Hi,

Check the Nfs services,
1. if needed try restart once and check .
2. check for the rpcstatd,rpclockd daemons.
3. Check rpcinfo -v
4. Check the connectivity between the NFS server and client.
5. Check for the /etc/exports file.
6. if needed execute exportfs -av

Thanks,
Analyst.

[Dave Olker]

Hi,

Which NFS patch are you running on this system?

How many threads are associated with the nfsktcpd process when it gets slow? This one process is just a parent where all the threads register their CPU time against, so this process is likely doing very little but the worker threads are busy.

When the server gets very busy you can try doing an:

# nfsstat -z

then wait a few minutes and do an:

# nfsstat -s

to see what kind of NFS activity the server is doing during the slow period.

Regards,

Dave