[Icecast] ices2, metadata, bumps and crashes

[Paul Martin]

On Tue, Aug 16, 2005 at 10:00:27PM +0200, Michael Smith wrote:
> On 8/16/05, Paul Martin <pm at nowster.zetnet.co.uk> wrote:

> > Caught it yesterday, as I was already running level 4:
> 
> This line:
> [2005-08-15  04:10:58] EROR stream/ices_instance_stream Send error: No
> error (Success)
> 
> is the one that catches my attention. Unfortunately, ices's error
> logging is somewhat sub-standard. Can you cross reference this against
> your icecast error.log to see if there's anything to indicate why this
> instance errored out (which it did - the 'Success' being printed is a
> problem with errno getting lost)?

The only significant info is here:

[2005-08-15  04:10:59] DBUG stats/modify_node_event update node subtype (Vorbis)
[2005-08-15  04:11:09] DBUG source/get_next_buffer last 1124075458, timeout 10, now 1124075469
[2005-08-15  04:11:09] WARN source/get_next_buffer Disconnecting source due to socket timeout
[2005-08-15  04:11:09] INFO source/source_shutdown Source "/medium.ogg" exiting
[2005-08-15  04:11:09] DBUG source/source_clear_source clearing source "/stereo44.ogg"
[2005-08-15  04:11:09] DBUG format-ogg/format_ogg_free_headers releasing header pages
[2005-08-15  04:11:09] DBUG format-ogg/free_ogg_codecs freeing codecs
[2005-08-15  04:11:09] DBUG format-vorbis/vorbis_codec_free freeing vorbis codec
[2005-08-15  04:11:09] DBUG source/source_free_source freeing source "/stereo44.ogg"
[2005-08-15  04:11:09] DBUG stats/modify_node_event update node sources (2)
[2005-08-15  04:11:09] DBUG stats/process_source_event delete source node /stereo44.ogg
[2005-08-15  04:11:22] DBUG slave/_slave_thread checking master stream list
[2005-08-15  04:13:23] DBUG slave/_slave_thread checking master stream list
[2005-08-15  04:15:23] DBUG slave/_slave_thread checking master stream list


> > Well, ogg123 definitely has its own problems too, probably a 2^31
> > rollover problem. (It spits out bursts of white noise after about 3
> > hours at 44.1kHz.)
> 
> I think I've noticed that one in the past, but I almost never use
> ogg123, so I haven't looked into it.
> 
> > 
> > The bump is noticeable. I suspect it's due to the encoder being
> > reinitialised in order to get a new serial number.
> 
> That cause a substantial jump in both cpu usage and bitrate for a
> moment, but if your client has any buffering at all (and most of them
> buffer quite a lot), then that shouldn't be a problem in any way. I'm
> curious as to why it is... it might just be the cpu usage causing the
> (hardware) audio buffers to get starved slightly - the application
> buffers (in ogg123 or other software) are more than big enough to
> cope.

The client I use most is on a lightly loaded 900MHz PIII. The server's 
an Athlon XP1800.

-- 
Paul Martin <pm at zetnet.net> (work)
  <pm at nowster.zetnet.co.uk> (home)