Hi,

I have a cluster running Centos5.3. I have login/user nodes that are nfs shared/automounted on my compute nodes. NFS is working properly with access/read/write.

My issue is that when one of my users runs his job, the number nfs daemon on the user/login node increases. From top, the nfsd daemon goes into either S or D. To me this indicates a lot of read/writes from the compute nodes to the login/user node.

I know the program the user is running, but is there a way to drill down to find out exactly what files from the program are causing the excessive r/w on the user/login node.

Thanks
mef

You could check it w/ lsof


Cheers,
Tink

Thanks, that is a good starting place.