Description of problem:

After a while, our nfs clients (using autofs) will stop working, and 
we will have stale handles all over the place, in most cases to have 
the affected people log out and then umount the directory in question 
and/or /etc/init.d/autof reload/restart will sometimes work, but 
after that the problem will come back very shortly.  This 
happens everytime when people want to start compiling (over nfs) and 
use vnc and such over nfs.  We are only a 300 people software company 
and have 3 fileservers (big ones), so I think this is abnormal.

This happens on all our 7.X boxes and redhat AS 2.1, kernel versions 
don't seem to matter, we are currently running the latest ones 
2.4.20-18.7 (smp where appliceble) and kernel-enterprise-2.4.9-e.25.
But some boxes are running older kernels.

This has serious problems for us.

Version-Release number of selected component (if applicable):
Ad 2.1
RH 7.3

How reproducible:

There seems to be very little rule as to how to make this happen, 
this happens all the 
time, when we are working over nfs.

Steps to Reproduce:
1.
2.
3.
  
Actual results:


Expected results:


Additional info: