On Tue, Nov 30, 2004 at 11:52:30AM -0800, Ephi Dror wrote:
> Hi all,
>
> I am running some tests using 3.0.9 and observed the following panics.
> I was also running also 3.0.6 and got same panics.


smbd is panicing in this case as it has detected internal corruption
in the locking tdb.

> Here are sample of the panics:
> ==============================
> [2004/11/30 11:24:52, 0] smbd/server.c:main(760)
> smbd version 3.0.9 started.
> Copyright Andrew Tridgell and the Samba Team 1992-2004
> [2004/11/30 11:24:52, 0] smbd/server.c:main(794)
> standard input is not a socket, assuming -D option
> [2004/11/30 11:25:31, 1] smbd/service.c:make_connection_snum(648)
> telaviv (10.123.4.132) connect to service ephi_test initially as user root
> (uid=0, gid=0) (pid 1717)
> [2004/11/30 11:25:32, 1] smbd/service.c:make_connection_snum(648)
> telaviv (10.123.4.132) connect to service ephi_test initially as user root
> (uid=0, gid=0) (pid 1717)
> [2004/11/30 11:25:32, 1] smbd/service.c:close_cnum(836)
> telaviv (10.123.4.132) closed connection to service ephi_test
> [2004/11/30 11:26:05, 1] smbd/service.c:make_connection_snum(648)
> paris (10.123.130.230) connect to service ephi_test initially as user root
> (uid=0, gid=0) (pid 2746)
> [2004/11/30 11:29:19, 0] smbd/oplock.c:request_oplock_break(1055)
> request_oplock_break: no response received to oplock break request to pid
> 2746 on port 32873 for dev = fd03, inode = 98351, file_id = 1335
> [2004/11/30 11:29:19, 0] smbd/open.cpen_mode_check(734)
> open_mode_check: exlusive oplock left by process 2746 after break ! For
> file test1/Issues.doc, dev = fd03, inode = 98351. Deleting it to continue...
> [2004/11/30 11:29:19, 0] smbd/open.cpen_mode_check(738)
> open_mode_check: Existent process 2746 left active oplock.
> [2004/11/30 11:29:36, 0] smbd/oplock.cplock_break(866)
> oplock_break: no break received from client within 30 seconds.
> oplock_break failed for file test1/Issues.doc (dev = fd03, inode = 98351,
> file_id = 1335).
> [2004/11/30 11:29:36, 0] smbd/oplock.cplock_break(911)
> oplock_break: client failure in oplock break in file test1/Issues.doc
> [2004/11/30 11:29:36, 0] tdb/tdbutil.c:tdb_log(725)
> tdb(/home/ephi/samba/var/locks/locking.tdb): tdb_lock failed on list 4
> ltype=1 (Resource deadlock avoided)


Ok, you're getting oplock break failures from the
client, and then EDEADLCK errors in the tdb fcntl
locking.

Can you reproduce this at will ? Can you send me the
Windows client tests you're using for this ?

I'd like to see the debug level 10 logs split out
into one log per active smbd process (use the %d
macro in the logname to do this). This might help
in showing the problem.

Thanks,

Jeremy.