Summary:ASTERISK-07556: chan_mgcp hangs for uknown reason
Reporter:dym dym (dym)Labels:
Date Opened:2006-08-19 04:52:29Date Closed:2011-06-07 14:02:41
Versions:Frequency of
Environment:Attachments:( 0) bug7760_backtrace.txt
( 1) bug7760_log.txt
( 2) bug7760_logfile.txt
( 3) bug7760_sourcemod.txt
Description:I have asterisk installation with hundred mgcp media gateways and a few sip phones. Sometimes I'm getting strange problem with asterisk when mgcp stops respond. In those situations all other parts of asterisk are working fine and f.e. SIP phones are working w/o any problem.


Media gateways: Terayon TA-102

O/S: Debian GNU/Linux 3.1, kernel: 2.4.27-3-686-smp

Asterisk version:
astro*CLI> show version
Asterisk SVN-branch-1.2-r40310M built by dym @ astro on a i686 running Linux on 2006-08-19 05:19:53 UTC

But the same problem I've got with asterisk-1.2.10

port = 2727
bindaddr =
disallow = all
allow = ulaw

;;;; Device 1
context = my_in
host = dynamic
defaultip =
callgroup = 0
pickupgroup = 0
nat = no
threewaycalling = no
transfer = no
callwaiting = yes
cancallforward = yes
canreinvite = no
singlepath = no
slowsequence = no
wcardep = aaln/*
dtmfmode = rfc2833
;;;; Line 1
callerid = "dym" <1301>
mailbox = 1301
line => aaln/1

All other devices looks the same way, coz they were generated by script from database.

root@astro:~# gdb /usr/sbin/asterisk 9156
[[[snipped useless backtrace]]]

Last message in log file:
[[[moved to file attachment]]]

And then once again "stopping retransmission..."

After that I didn't see any record about mgcp in logfile, chan_mgcp got frozen. Before this I have problem with transfer (when you pressing flash button while the call and then hanging up, symptoms were the same), so I disabled transfers. Seems users have found another method for hanging asterisk.

If you'll have time and want to get more information please ask.
Comments:By: Tilghman Lesher (tilghman) 2006-08-20 10:49:26

Please recompile Asterisk with optimizations turned off (i.e. make clean dont-optimize) and redo your backtrace, then UPLOAD your backtrace to the file
upload section.

Please do not paste debugging output into the bugnotes.

By: dym dym (dym) 2006-08-21 02:24:44

I've recompiled asterisk w/o optimization and just uploaded the file w/ backtrace. If you need I can attach logfile too.

By: Serge Vecher (serge-v) 2006-08-21 10:32:10

1) please do upload logfile as attachment.
2) what modifications were done to the source?

By: dym dym (dym) 2006-08-21 23:47:26

1) Ok, uploaded.
2) Look at file bug7760_sourcemod.txt. I just turned off some debug output, because it was just flooding log. This is the only change.

By: stephen dredge (stephen_dredge) 2006-09-06 21:16:53

This problem could be because when slow sequence is not set, each sub channel has it's own queue for media related commands ( create, modify, delete connection). The driver only attempts to match a response  to one of these commands in the current subchannels queue. Thus if the current subchannel has changed in the mean time the response is not matched with the request and the queue will stall.

Setting slow sequence true might help.

By: dym dym (dym) 2006-09-14 00:19:25

Thanks, stephen_dredge.
I didn't see this problem last 5 days after setting slow sequence true. So seems subchannel queues were my unknown reason problem :)

PS: asterisk continue hanging when I interrupt transfer:
"""Before this I have problem with transfer (when you pressing flash button while the call and then hanging up, symptoms were the same), so I disabled transfers."""

By: litw (litw) 2006-09-25 08:28:29

did you try apply that patch ?

I have 4 heavy loaded mgcp GWs - 3 Dlink DG104 and one DVG-1024M
if I use not patched channel driver asterisk locks within 30 min.

By: dym dym (dym) 2006-09-25 08:49:50

Yes, I tried that patch. But it did not solve my problem.

By: jmls (jmls) 2006-11-01 13:03:34.000-0600

anything else we can try on this ?

By: Anton Vazir (vazir) 2007-02-10 02:29:05.000-0600

dym, was the issue resolved?

By: dym dym (dym) 2007-03-05 04:40:05.000-0600

Sorry for late responce.

2Vazir: no, problem was not solved. Asterisk is working w/ disabled transfers, so it's not hanging anymore.

By: Jason Parker (jparker) 2007-07-03 14:14:24

There are very few (if any) developers that are even able to test using MGCP.  Without a proper backtrace, and without anybody familiar enough with chan_mgcp to debug/fix the problem, there is very little we can do here.

I'm going to have to close this, unless/until a proper backtrace can be generated.