Icecast Streaming Media Server

[Douglas]

Hello folks, this is my first post, although I have spent a bit of time reading many others.

I am running a fairly high traffic icecast server (maybe 1800 listeners or so). I am using 2.3.2-kh1.

I am using FreeBSD-amd64 SMP

here is a backtrace

#0 0x000000080169ba64 in pthread_mutex_lock () from /lib/libthr.so.3
#1 0x000000000040cdd1 in start_relay_stream (arg=Variable "arg" is not available.
) at slave.c:431
#2 0x000000000042281a in _start_routine (arg=0x9f4c60) at thread.c:584
#3 0x0000000801697a27 in pthread_getprio () from /lib/libthr.so.3
#4 0x00007ffffe757000 in ?? ()
Cannot access memory at address 0x7ffffe7d700

some higlights from messages:

Sep 22 01:33:09 stream2 kernel: pid 1755 (icecast), uid 400: exited on signal 11
Sep 22 01:33:19 stream2 kernel: pid 15201 (icecast), uid 400: exited on signal 11
Sep 23 09:46:08 stream2 kernel: pid 710 (icecast), uid 400: exited on signal 11
Sep 23 10:45:28 stream2 kernel: pid 84833 (icecast), uid 400: exited on signal 11
Sep 23 10:58:18 stream2 kernel: pid 95805 (icecast), uid 400: exited on signal 11
Sep 23 11:04:53 stream2 kernel: pid 97495 (icecast), uid 400: exited on signal 11
Sep 23 12:13:01 stream2 kernel: pid 701 (icecast), uid 400: exited on signal 11
Sep 23 13:47:29 stream2 kernel: pid 10894 (icecast), uid 400: exited on signal 11 (core dumped)
Sep 23 13:48:13 stream2 kernel: pid 46863 (icecast), uid 400: exited on signal 11 (core dumped)
Sep 23 13:49:29 stream2 kernel: pid 46976 (icecast), uid 400: exited on signal 11 (core dumped)

(I will gladly include any other information, logs et cetera if needed)

I haven't determined how to reproduce this but it seems to happen a) under high traffic (over 100 mbit) b) every few hours or less.

I am also using varnishd for some html elements and redirections based on URL but I don't _think_ that is related.

Thanks. I would like to figure this out so my users don't disconnect sporadically.

[karlH]

haven't had a crash report in a while. Can you make sure that the build was built using "make debug" as optimizations can make debug difficult. If we assume that the problem is as indicated then it's at the point of starting a relay, so maybe a look at the error log is required. Email the details of the xml and logs so I can look over the specifics.

karl.

[Douglas]

This is the backtrace from a debug compile.

#0 0x00000008016a6a64 in pthread_mutex_lock () from /lib/libthr.so.3
#1 0x000000000042b945 in _mutex_lock (mutex=0x100) at thread.c:650
#2 0x000000000042b3dc in thread_mutex_lock_c (mutex=0x100, line=431,
file=0x4330b3 "slave.c") at thread.c:374
#3 0x000000000040f96d in start_relay_stream (arg=0x864700) at slave.c:431
#4 0x000000000042b7d6 in _start_routine (arg=0x806500) at thread.c:584
#5 0x00000008016a2a27 in pthread_getprio () from /lib/libthr.so.3
#6 0x00007ffffcb1f000 in ?? ()
Cannot access memory at address 0x7ffffcb9f000

I am happy to dig up any info or run any tests that might help. Thanks again.

[Douglas]

I applied a patch Karl wrote and the server has not crashed since. It works very well.

Thank you Karl. I wanted to give things a few days, and report back.

Seems to work great!

[karlH]

glad to hear it, it will be in kh2 which I'll release when I get some feedback on something else that is not crash case. Just a note for people, it's obviously no so easy to trigger but it could trigger if configured as a slave.

karl.