Hi Chris,
If it takes a win2k client to make this happen, and you can reproduce,
with a specific win2k client machine, do the following to get a debug of
what that client is sending to the samba server during the 'explosion':
in smb.conf global section:
log level = 0
log file = /usr/local/samba/var/log.%m
include /usr/local/samba/var/debug.%m
and create a file named /usr/local/samba/var/debug.win2k-machine-name that
you are using for the reproduction.
In that file have the single line
log level = 10
This will cause all your other clients to function without the overhead of
debugging, but when the win2k machine attaches, it will read in the
debug.machinename, set the logging level to 10, and then when you see your
'explosion' you should have a
/usr/local/samba/var/log.win2k-machine-name to
look at and see IF and WHY it is trying to reconnect so often - this will
tell us if the win2k machine is actually flooding samba with connection
requests, or if we have a problem in the allocation part of tdb that is
triggering off of some field in the request from the client....
Hope this helps,
Don
-----Original Message-----
From: root [mailto:lists@gecko.math.ualberta.ca]
Sent: Tuesday, March 20, 2001 10:54 PM
To: samba@lists.samba.org
Subject: Exploding connections
i've managed to catch the exploding connections database in the act. i was
able to attach gdb to the smbd while it was going crazy:
(gdb) bt
#0 0xf55da in tdb_expand ()
#1 0xf5787 in tdb_allocate ()
#2 0xf6554 in tdb_store ()
#3 0x6534 in claim_connection ()
#4 0x406a2 in make_connection ()
#5 0x1e105 in reply_tcon_and_X ()
#6 0x3e256 in switch_message ()
#7 0x3e8c7 in chain_reply ()
#8 0x1ffc3 in reply_sesssetup_and_X ()
#9 0x3e256 in switch_message ()
#10 0x3e2f2 in construct_reply ()
#11 0x3e5fe in process_smb ()
#12 0x3ef1f in smbd_process ()
#13 0x3479 in main ()
it seems that the problem has something to do with win2k systems: i can
have a whole bunch of win98 and NT sessions open to the share, but win2k
reliably causes this problem. the problem is that connections.tdb is
growing about 1MB/s - almost as if 2k were flooding the server. i can
reproduce this with both SAMBA_2_2 and HEAD, from a cvs update no more
than an hour old.
what other information should i provide to help track this bugger down?
CK
--
Chris Kuethe: System Administrator - U of A Math Dept.
cell: 780.903.9475 office: CAB553, 780.492.1704
wargames@{edmc.net,hackcanada.com} ckuethe@[[gecko.]math.]ualberta.ca
Opinions expressed herein are solely the responsibility of
the author. And the author wouldn't have it any other way.
--
To unsubscribe from this list go to the following URL and read the
instructions: http://lists.samba.org/mailman/listinfo/samba