Summary:ASTERISK-09009: Random Crashes on IAX2 trunk
Reporter:Fabricio Machado (fmachado)Labels:
Date Opened:2007-03-14 12:31:03Date Closed:2007-06-30 09:20:08
Versions:Frequency of
Environment:Attachments:( 0) bt_full.txt
( 1) debug.txt.gz
( 2) thread_apply_all_bt.txt

 In our production environment, we have two asterisk servers with an IAX2 trunking between them.
 Unfortunately, we got randomly segmentation faults since version 1.4.0.
 This segmentation faults occurs when a call is made through IAX2 trunk.

Our scenario:
extensions prococol: IAX2
softphone: IDEFISK
softphone codec: ALAW
asterisk version: SVN 58512 (both servers)
* Both are compiled with DONT_OPTIMIZE and DEBUG_THREADS options.

Hardware Description:

Server 'A':
-Dell PowerEdge 1800
-Dual Intel(R) Xeon(TM) CPU 3.00GHz Coreduo
-Digium TE110P card
-Digium TDM2400 (with 6 FXS modules)
-Digium TDM400P (with 4 FXO modules)

Server 'B':
-Dell PowerEdge 430SC
-Intel P4 2.4Ghz (hyperthreading enabled)
-Digium TE110P card
-Digium TDM400P (with 4 FXS modules)
Comments:By: Fabricio Machado (fmachado) 2007-03-14 13:24:23

Attachment descriptions:

bt_full.txt: Contains 'bt full' output of gdb command on 'Server B' coredump file.

thread_apply_all_bt.txt: Contains 'thread apply all bt' output of gdb command on 'Server B' coredump file.

debug.txt.gz: Contains last 2 or 3 minutes of asterisk operation on 'Server B' before crashing (same crash of coredump above)

By: Joshua C. Colp (jcolp) 2007-03-15 13:21:30

Please try 1.4 as of revision 58923 or trunk as of revision 58924. Thanks!

By: Fabricio Machado (fmachado) 2007-03-15 13:44:36

OK, we will try this update at this night. (at 6:30 pm / GMT -3)
Tomorrow, we'll give a feedback.

By: Fabricio Machado (fmachado) 2007-03-16 10:54:07

After update, it seems ok.
But is better to wait 3 days of full operational state to close this case.
On the next tuesday (March,20th) we will got these 3 days of full operational state. So, I'll give a final feedback.

By: Serge Vecher (serge-v) 2007-03-16 11:19:38

thanks for reporting. Let's close the issue and if the problem does resurface later on, please reopen the issue with a new backtrace.