View previous topic :: View next topic |
Author |
Message |
TheFlyingBear
Joined: 08 Dec 2010 Posts: 15 Location: France
|
Posted: Tue Dec 14, 2010 8:56 am Post subject: Icecast 2.3.2-kh28 - trap divide error |
|
|
Hello,
Tonight, one of my icecast server "died". The server was really far from what we may call loaded (a few connections to the server, low load, less than 200MB of memory used).
I found this error in my /var/log/messages:
Quote: |
Dec 14 02:12:21 icecast21 kernel: icecast[6674] trap divide error ip:423586 sp:7fa272693bf0 error:0 in icecast[400000+3d000] |
The server is running Gentoo Linux, with a 2.6.34-gentoo-r12 kernel and icecast 2.3.2-kh28.
Don't hesitate to tell me if you need more informations. _________________ Guillaume |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
karlH Code Warrior
![Code Warrior Code Warrior](templates/subRebel/images/ranks/rank_dev.gif)
Joined: 13 Jun 2005 Posts: 5476 Location: UK
|
Posted: Tue Dec 14, 2010 10:01 pm Post subject: |
|
|
I would need a backtrace of the fault. This is probably a divide by zero error but without knowing the point of failure, it's hard to solve. If you can get the kh28.4 update and build with "make debug" then run it so a core file can be produced (eg ulimit -c unlimited), then we can check where the failure is.
karl. |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
TheFlyingBear
Joined: 08 Dec 2010 Posts: 15 Location: France
|
Posted: Tue Dec 14, 2010 10:18 pm Post subject: |
|
|
OK. I'll do this tomorrow and wait for a crash. _________________ Guillaume |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
karlH Code Warrior
![Code Warrior Code Warrior](templates/subRebel/images/ranks/rank_dev.gif)
Joined: 13 Jun 2005 Posts: 5476 Location: UK
|
Posted: Wed Dec 15, 2010 1:10 am Post subject: |
|
|
BTW, you can always simulate a crash with killall -SEGV icecast, that will prove whether the core file generation is set up correctly.
karl. |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
TheFlyingBear
Joined: 08 Dec 2010 Posts: 15 Location: France
|
Posted: Fri Dec 17, 2010 3:29 pm Post subject: |
|
|
I installed a debug version of Icecast, and allowed core dump on the system.
I'm waiting for the next occurence of this problem.
I have 2 new strange problems I'll create new threads for them... _________________ Guillaume |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
TheFlyingBear
Joined: 08 Dec 2010 Posts: 15 Location: France
|
Posted: Tue Dec 21, 2010 4:26 pm Post subject: |
|
|
I could wait for icecast to dump a core: it's not enabled in my kernel
I will try to upgrade my kernel asap. _________________ Guillaume |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
TheFlyingBear
Joined: 08 Dec 2010 Posts: 15 Location: France
|
Posted: Thu Dec 23, 2010 4:48 pm Post subject: |
|
|
I'm finally able to generate core on my servers, but I can't do it with the icecast process.
I'm a using a version compiled with "make debug", the /proc/`pidof icecast`/limits says that there is no limit on core size, the icecast user can write in the directory where I want to put my cores, but, neither a kill -SEGV nor the trap divide error can produce a core
Can any other information be usefull to you to track this error (error.log, icecast.xml, ...)? _________________ Guillaume |
|
Back to top |
|
![](templates/subRebel/images/spacer.gif) |
|