[icecast] ices stops connecting to server after 24hours

Karl Heyes karl at xiph.org
Sun Oct 12 13:57:20 UTC 2003



On Sun, 2003-10-12 at 08:53, Matt Trim wrote:
> Hi Karl,
> 
> I just tried the latest - same problem :(
> 
> One of the lines in the log file seems interesting - see the (Out of memory)
> error..  After that appears it seems to never want to connect after that.
> My box has plenty of memory - ices seems to only use 30 megs of memory and
> around 29% cpu.  Memory leak maybe?

The out of memory message is a libshout error, it doesn't necessarily
correspond with the virtual memory of the process, as there are several
safeguards against runaway memory allocations.  The 29% CPU indicates
that the instances are still doing the encoding, and as file writes are
ok then it looks more like a libshout issue.

> [2003-10-12  17:29:40] EROR om_shout/output_ogg_shout Failed to write to
> xstream1.com:8000/xxxx.ogg (Socket error)
> [2003-10-12  17:29:40] DBUG om_shout/_output_connection_close closed shout
> connection
> [2003-10-12  17:29:45] DBUG om_shout/check_shout_connected Time we started
> stream on xstream1.com:8000/xxxx.ogg
> [2003-10-12  17:29:46] EROR om_shout/check_shout_connected Failed to connect
> to xstream1.com:8000/xxxx.ogg (Out of memory)
> [2003-10-12  17:29:51] DBUG om_shout/check_shout_connected Time we started
> stream on xstream1.com:8000/xxxx.ogg
> [2003-10-12  17:29:51] EROR om_shout/check_shout_connected Failed to connect
> to xstream1.com:8000/xxxx.ogg (Cannot set para
> meter while connected)
> [2003-10-12  17:29:56] DBUG om_shout/check_shout_connected Time we started
> stream on xstream1.com:8000/xxxx.ogg
> [2003-10-12  17:29:56] EROR om_shout/check_shout_connected Failed to connect
> to xstream1.com:8000/xxxx.ogg (Cannot set para
> meter while connected)
> 
> Nothing odd is going on at the 24 hour mark - I do have a dual processor
> machine which I believe can cause problems with some code?

Usually with this sort of error, something occurs (eg cron job) that
triggers something like a CPU/network saturation/outage, that looks to
be affecting libshout.  It's unlikely to be a multiple CPU issue.

> I will email you the log output of ices as it's too long to post here (I'll
> gzip it for you)

I've received it, thanks

karl.

<p>--- >8 ----
List archives:  http://www.xiph.org/archives/
icecast project homepage: http://www.icecast.org/
To unsubscribe from this list, send a message to 'icecast-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.



More information about the Icecast mailing list