NFS problem - not responding

[Coolmar]

Hi,

I keep getting the following error message:

NFS server serverb not responding still trying
NFS server serverb not responding still trying
NFS server serverb not responding still trying
NFS server serverb not responding still trying

I have a whole bunch of roque bdf processes that I can't kill even with a -9. I found in the /etc/mnttab that it is trying to nfs mount a server that is dead. I tried to remove the server from the /etc/mnttab but it keeps putting it back in. Is there a way that I can get it to stop "looking" for serverb and a way to delete the line from the /etc/mnttab and keep it out.

Thanks!

[Steven E. Protter · Accepted solution]

The box getting the messages may need to be rebooted.

What probably happened is that there were a bunch of open files and then network connectivity was lost. If its lost for two long, it can be such a pain to clean up, booting the client(the server/box getting this error) is easier.

Check that you have complete network connectivity to serverb.

Also, make sure there have been no recent changes to the serverb:/etc/exports file that might block access to the machine getting this error.

Before taking any action, make sure networking is fully restored.

SEP

[Devender Khatana]

Hi,

Removing from mnttab will not help because it is created automatically for all mounted file systems. A better approach will be to remove it from /etc/fstab if it is there & from /etc/auto.direct & /etc/auto_master if it is NFS automounter configuration.

This can be avoided by stoping NFS client also.

#/sbin/init.d/nfs.client stop

If the NFS server is down it could even require reboot for the client as well.

HTH,
Devender

[Mel Burslan]

The simple answer to this question is NO. As mnttab is created/updated dynamically, you will not have much luck.

And you will not be able to kill the rogue bdf processes until the next time you reboot more than likely.

I have never tried this but, if your nfs mounts are against a name, not an IP address directly, can you dummy up some identical NFS mount points on a server where NFS server is active, and on your local host's /etc/hosts file (hope your nsswitch.conf is configured to check this for address resolution) enter the down host's name mapping to the dummy host you just created, and let your server go check this new mount points, at which case, your bdf processes may have achance to complete and die.

But more than likely, you are looking at a reboot if you are in need of getting rid of thise desperately.

[V. Nyga]

Hi,

check also /etc/fstab
and umount all directories from severb

Volkmar

[Pete Randall]

Sally,

Your rogue bdf processes will not go away until the I/O request that they are waiting on is satisfied. The same is true for any other process that is trying to access serverb. The only thing that MAY work is to stop and restart the nfs client process on the client, otherwise it will require a re-boot.


Pete

[Rita C Workman]

You can't edit this out of /etc/mnttab.
Check in your /etc/fstab and remove it from trying to mount on a reboot.

I had a system that had headaches for awhile on some NFS mount used by an application.
Cleaning up hung NFS mounts is a bear, I could never get quite get them cleaned up, cause I got left with too much garbage/zombies and mess.
I found I had to reboot the box.

So edit and ensure that mount will not happen either in a pkg.cntl file or in /etc/fstab.

Rgrds,
Rita

[Coolmar]

Thanks all...looks like I will be rebooting it tonight.