samba - Jul 2004 - 3.0.4 on Solaris 8 - fcntl(F_SETLK64)

We upgraded from 2.2.5 to 3.0.4 last week and ever since have been running
into a deadlock condition that causes clients to hang for extended
periods; all trying to get a write lock on locking.tdb.  I have been
dilligently googling to find the solution and have thus far been
unsuccessful.  My kernel rev is 108528-29, so I think I'm ok there.  Does
anyone have any ideas?

All I can think of is that I have a gcc install that I used to build samba
from before the kernel patch and gcc "fixed" a crucial header that
changed
with the 108528-19 patch?

I had to revert today to keep our users from killing me ;-).  Please let
me know what technical info I can provide to track this down.  I'm pretty
handy with gdb, truss, etc.  Thanks.

-- 
Brian Ford
Senior Realtime Software Engineer
VITAL - Visual Simulation Systems
FlightSafety International
the best safety device in any aircraft is a well-trained pilot...

On Wed, Jul 28, 2004 at 11:52:03AM -0500, Brian Ford
wrote:
> We upgraded from 2.2.5 to 3.0.4 last week and ever since have been running
> into a deadlock condition that causes clients to hang for extended
> periods; all trying to get a write lock on locking.tdb.  I have been
> dilligently googling to find the solution and have thus far been
> unsuccessful.  My kernel rev is 108528-29, so I think I'm ok there. 
Does
> anyone have any ideas?
> 
> All I can think of is that I have a gcc install that I used to build samba
> from before the kernel patch and gcc "fixed" a crucial header
that changed
> with the 108528-19 patch?
> 
> I had to revert today to keep our users from killing me ;-).  Please let
> me know what technical info I can provide to track this down.  I'm
pretty
> handy with gdb, truss, etc.  Thanks.
Can you put this back on a test server then start looking at the truss
output to see what it is waiting on. This does sound very much like the
problem with the thundering herd problem with solaris fcntl locks...

Jeremy.