Hung smbd processes and failing server

[Sid VanHouten CT90]

Running samba 2.06 on Solaris 2.6 and CIFS/9000 on HPUX11.  Accessing from 
NT 4.0 Terminal server.

We are seeing the following symptoms.  Can anyone offer any diagnostic advice...

Users will connect to samba servers and be working along ok.  Without warning,
the connection starts to act flaky.  e.g. can't save documents, file system
browsing becomes very erratic.

A quick "ps | grep smbd" indicates that these users are piling up lots of smbd
processes that do not go away short of explicitly killing them or restarting
samba.  I suspect the users are trying many logoff/logon cycles to try to 
clear up the connection problems.

There was only one log message that I was uncertain about...
[2000/04/14 07:54:08, 0] smbd/oplock.c:request_oplock_break(1202)
  request_oplock_break: no response received to oplock break request to pid 
22262 on port 47994 for dev
 = 2b80988, inode = 166829

I don't know what this message is telling me, and it's not clear to me if it 
might be related to the problems.

We have taken a look at network monitor info and don't see any significant 
issues.

Any thoughts?

Sid VanHouten                   svanhout at gmr.com