[OpenSIPS-Devel] [ opensips-Bugs-2818650 ] signo=Could not find the frame base for "sig_alarm_abort"

Fri Jul 10 15:07:40 CEST 2009

Bugs item #2818650, was opened at 2009-07-08 19:49
Message generated for change (Comment added) made by tomvogt
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1086410&aid=2818650&group_id=232389

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: core
Group: 1.4.x
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Thomas V (tomvogt)
Assigned to: Bogdan-Andrei Iancu (bogdan_iancu)
Summary: signo=Could not find the frame base for "sig_alarm_abort"

Initial Comment:
opensips 1.4.5-notls, FreeBSD 7.0 (i386). opensips core dumps after a few hours with:

#0  0x28268017 in kill () from /lib/libc.so.7
#1  0x28267f76 in raise () from /lib/libc.so.7
#2  0x28266b8a in abort () from /lib/libc.so.7
#3  0x08067ee3 in sig_alarm_abort (signo=Could not find the frame base for "sig_alarm_abort".
) at main.c:423
#4  <signal handler called>
#5  fm_status (qm=0x287fe000) at mem/f_malloc.c:512
#6  0x080679ed in cleanup (show_status=1) at main.c:356
#7  0x08068795 in handle_sigs () at main.c:524
#8  0x0806c969 in main (argc=3, argv=0xbfbfec58) at main.c:866

any idea?

----------------------------------------------------------------------

Comment By: Thomas V (tomvogt)
Date: 2009-07-10 15:07

Message:
It crashes also with more Ram. Looks like memory is not the case. I do not
have more information.

With an older opensips 1.4.2-notls the system always crashed with:
SIP/2.0 180 Ringing\r\nVia: SIP/2.0/UDP
212.xxx.xxx.xxx;branch=z9hG4bK0622.99552112.0\r\nVia: SIP/2.0/UDP 
85.xxx.xxx.xxx4:5060;received=85.xxx.xxx.xxx;rport=5060;branch=z9hG4bKc3ba5c8f50bc782082391762e9a2fa9e\r\nF"...,
len=503, 
  rcv_info=0xbfbfeb14

A ngrep when 1.4.2 was running showed:
U 85.xxx.xxx.xxx:5060 -> 212.xxx.xxx.xxx:5060
INVITE sip:xxxxxxxx at 212.xxx.xxx.xxx SIP/2.0.
Via: SIP/2.0/UDP
85.xxx.xxx.xxx4:5060;rport;branch=z9hG4bK3759d84a0bea3edc47d08b9ec1cd2979.
To: <sip:xxxxxxx at 212.xxx.xxx.xxx>.
From: <sip:xxxxxx5 at 212.xxx.xxx.xxx>;tag=AI60E7E8F4CC7D6BD2.
Call-ID: AI1E214F5EDC57C05C at 192.168.1.240.
CSeq: 1 INVITE.
Max-Forwards: 70.
Contact: <sip:xxxxxxx at 85.xxx.xxx.xxx;line=AIDF22AB43A35036D9>.
Accept: application/sdp.
Allow: ACK,BYE,CANCEL,INVITE,NOTIFY,OPTIONS,REFER.
P-Preferred-Identity: <sip:xxxxxxxxx at 212.xxx.xxx.xxx5>.
Privacy: none.
User-Agent: Aastra Intelligate.
Content-Type: application/sdp.
Content-Length: 248.
.

.

.

.

.

.

.

.

.

#
I 212.xxx.xxx.xxx -> 85.xxx.xxx.xxx 3:3
...E(..k...;.
Uv...e.}~ ....;l
#

We installed Revision: 5049. After that we had the sig_alarm_abort issue.
So we updated to 1.4.5-notls. Nothing changed, opensips still crashes. 

----------------------------------------------------------------------

Comment By: Thomas V (tomvogt)
Date: 2009-07-10 11:13

Message:
Hi Bogdan

I'm not sure but it looks like a memory leak. I upgraded the system to
FreeBSD 7.2 and added an additional 1 GB memory (total 2gb). After 48h
opensips reserved at least 1gb (1gb is for postgresql). We do not allow to
use swap on the system. If the system has less than 50mb left, opensips
crashes

Log:

Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8408]:
INFO:core:handle_sigs: child process 8421 exited by a signal 11
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8408]:
INFO:core:handle_sigs: core was generated
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8408]:
INFO:core:handle_sigs: terminating due to SIGCHLD
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8439]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8435]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8434]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8433]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8432]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8427]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8426]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8425]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8424]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8423]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8420]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8420]:
CRITICAL:core:fm_status: different free frag. count: 15!=37 for hash   5
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8419]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8431]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8430]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8429]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8428]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:58:59 sip02 /usr/local/sbin/opensips[8450]: INFO:core:sig_usr:
signal 15 received
Jul 10 08:59:59 sip02 /usr/local/sbin/opensips[8408]:
CRITICAL:core:sig_alarm_abort: BUG - shutdown timeout triggered, dying...

coredump:

(gdb) where
#0  0x2826de67 in kill () from /lib/libc.so.7
#1  0x2826ddc6 in raise () from /lib/libc.so.7
#2  0x2826c9da in abort () from /lib/libc.so.7
#3  0x08067ee3 in sig_alarm_abort (signo=14) at main.c:422
#4  <signal handler called>
#5  0x080daf8b in fm_status (qm=0x28807000) at mem/f_malloc.c:530
#6  0x080679ed in cleanup (show_status=1) at fastlock.h:181
#7  0x08068795 in handle_sigs () at main.c:523
#8  0x0806c969 in main (argc=3, argv=0xbfbfec7c) at main.c:860
(gdb) frame 6
#6  0x080679ed in cleanup (show_status=1) at fastlock.h:181
181	fastlock.h: No such file or directory.
	in fastlock.h

----------------------------------------------------------------------

Comment By: Bogdan-Andrei Iancu (bogdan_iancu)
Date: 2009-07-09 09:54

Message:
Hi Thomas,

the core is actually showing an unsuccessful shutdown (see the cleanup()
on frame 6). The final coredump was generated because the shutdown took
longer than 20 seconds (and there is an ALARM signal to prevent blocking
during the shutdown). 
So the question is: in logs, do you have any reports related to the cause
for the initial shutdown ?

Regards,
Bogdan 

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1086410&aid=2818650&group_id=232389