[Home]

Summary:ASTERISK-10970: safe_asterisk fails to restart asterisk after crash
Reporter:Matthew M. Boedicker (mmb)Labels:
Date Opened:2007-12-04 09:12:27.000-0600Date Closed:2011-06-07 14:03:06
Priority:MajorRegression?No
Status:Closed/CompleteComponents:General
Versions:Frequency of
Occurrence
Related
Issues:
Environment:Attachments:
Description:If safe_asterisk is started from a terminal, it works as expected. If it is started from the init script, it will not restart the asterisk process. I believe this is due to safe_asterisk writing to stderr when it is not being redirected anywhere. The init script that calls safe_asterisk should redirect stderr because I don't think the daemon function does.

tested on CentOS 4.5

****** ADDITIONAL INFORMATION ******

Non-working case (safe_asterisk started from init script):

[root@box ~]# service asterisk start
Starting asterisk:                                         [  OK  ]
[root@box ~]# ps auxgww | grep aster
root     18618  0.0  0.0  6160  528 pts/2    S    10:00   0:00 /bin/sh /usr/sbin/safe_asterisk
root     18662  6.5  0.3 20328 7676 pts/2    Sl   10:00   0:00 /usr/sbin/asterisk -vvvg -c
root     18769  0.0  0.0  4164  628 pts/2    S+   10:00   0:00 grep aster
[root@box ~]# kill -11 18662
[root@box ~]# ps auxgww | grep aster
root     18797  0.0  0.0  5504  628 pts/2    S+   10:00   0:00 grep aster
[root@box ~]#

Working case (safe_asterisk started from terminal):

[root@box ~]# /usr/sbin/safe_asterisk
[root@box ~]# ps auxgww | grep aster
root     19156  0.0  0.0  4864  524 pts/2    S    10:02   0:00 /bin/sh /usr/sbin/safe_asterisk
root     19196  7.7  0.3 20064 7680 pts/2    Sl   10:02   0:00 /usr/sbin/asterisk -vvvg -c
root     19299  0.0  0.0  3880  628 pts/2    S+   10:02   0:00 grep aster
[root@box ~]# kill -11 19196
[root@box ~]# /usr/sbin/safe_asterisk: line 111: 19196 Segmentation fault      (core dumped) nice -n $PRIORITY ${ASTSBINDIR}/asterisk ${CLIARGS} ${ASTARGS} >&/dev/${TTY} </dev/${TTY}
Asterisk ended with exit status 139
Asterisk exited on signal 11.
Automatically restarting Asterisk.
mpg123: no process killed

[root@box ~]# ps auxgww | grep aster
root     19156  0.0  0.0  4872  680 pts/2    S    10:02   0:00 /bin/sh /usr/sbin/safe_asterisk
root     19363  2.1  0.3 20256 7680 pts/2    Sl   10:02   0:00 /usr/sbin/asterisk -vvvg -c
root     19466  0.0  0.0  5328  628 pts/2    S+   10:02   0:00 grep aster
[root@box ~]#
Comments:By: Phoebe Anderson (phoebe) 2007-12-05 08:10:57.000-0600

Are you using something like bootsplash?  If so, try launching safe_asterisk using "setsid" from your init script.
By: Matthew M. Boedicker (mmb) 2007-12-05 08:24:43.000-0600

Not using anything like bootsplash. What starts asterisk is:

init script -> daemon function -> initlog -> safe_asterisk -> nice -> asterisk
By: Terry Wilson (twilson) 2007-12-12 14:17:17.000-0600

Last I checked, redhat-based distros didn't do this correctly.  Problem (if it is still the same) was how the daemon() function for initscripts calls safe_asterisk.  Seems like I had to remove initlog from call to get it to work right.  Will put together a test box and see if I can find a fix.


By: Terry Wilson (twilson) 2007-12-12 14:23:34.000-0600

Actually it was initlog -c changed to initlog -r:

http://bugs.digium.com/view.php?id=901

Hopefully can find a solution that works with initlog -c ... of course, Fedora 8 has initlog as deprecated and the script works there, so eventually this won't be a problem.
By: Matthew M. Boedicker (mmb) 2007-12-12 15:39:06.000-0600

A temporary fix that seemed to work, however this bypasses initlog competely:

In /etc/init.d/asterisk:

Change:

daemon $DAEMON $ASTARGS

to:

nohup $DAEMON $ASTARGS
By: Terry Wilson (twilson) 2007-12-12 19:10:11.000-0600

also Centos 5.1 works fine (no initlog).


By: Tilghman Lesher (tilghman) 2008-01-07 15:07:03.000-0600

Okay, I think this is clearly a distroy specific issue, that has been fixed (or will be fixed) in later versions, so no fix needs to go into Asterisk.