I'm running on a suse 6.0 system with a 2.0.26 kernel.  I have tried the
2.2.2 kernel and I still get the same problem.  We have about 100+ nodes
on the network all filesystems besides system filesystems are
automounted on every machine.  Most of the machines have about 2-6
filesystems each and are of varying OS(AIX IRIX HPUX SUNOS Digital
UNIX).  We have several Linux machines running autofs.  We develop
software for all these platforms.  

Now the problem.  One user is trying to compile on several platforms at
once.  This causes a lot of nfs mounts and activity.  The compiles work
fine on a few machines and then start to fail.  He notices that several
of his source files are corrupt, they have binary junk at the end of
them.  The machine that it dies on usually has NFS timeout errors in
it's logs.  If he compiles one machine at a time they all work fine.

Is this a kernel tuning problem?  Should I move his machine over to
100Mbit to maybe relieve network congestion?  Is this a known problem?

What am I doing wrong?

Thanks
Drew

Reply via email to