nfs file stale handle
 

Hi,

on my cluster I have frequently problems with nfs... I get an error like:
Stale NFS file handle

This happens mostly during write out of an simulation or other access to the files (e.g. mapFields).
Unfortunately, the simple procedures explained here (http://www.cyberciti.biz/tips/nfs-st...-solution.html) are not
working always... so according to the statement of
Daemeon from January 29, 2009:
"When you force unmount/remount and still get the error, you have some
program that is probably (?) not following proper file handle semantics
about closing files.", there might be a problem in openfoam!?

This happens with 1.5-dev,1.6.x and 1.7.x. The cluster has OpenSUSE 11.2 installed.
The workstation is running with opensuse 11.1 and ubuntu lucid, which share the data directories using nfs.

Does anyone have similar problems and could have an advice, what I could check or adjust (maybe there are some nfs settings making it a bit more 'stable')!?

Regards!
Fabian

I don't think this is an OpenFOAM problem. I've only ever had problems with stale file handles eg. if I move a directory in one ssh session while it's currently the pwd in another session. Are you the administrator of the cluster? Maybe someone is changing something while you are working?

Hello,

thanks for your reply! I think so as well, that OpenFOAM is not the reason for this. Do you have a hint, what I am doing wrong in the setup of the cluster? Maybe the installed kernel 2.6.31.8-0.1 is the bad guy or needs some adjustments!? Or does there exist something different than nfs to share directories among machines?
Would be nice, if you can help!

Fabian

I'm sorry, I never set NFS up myself, so I can't really help with that. Hopefully someone more experienced in setting up a cluster stumbles on this thread. Another idea is taking the question to another forum with more network specialists, eg. http://serverfault.com/

- Anton

Thanks, I will check the other forum...