Icecast - Feb 2005 - Re: Severe problem with Icecast2 and streamTranscoder

Dennis Speekenbrink wrote:
> Does your server have enough continuous throughput to keep up?
I know it's not related to bandwidth/CPU/memory issues, at least not
directly.  Consider the following:

I've been running tests where I've set up more than one instance of
streamTranscoder - all running on the same machine as the Icecast server
and with identical config files - to transcode test.ogg into, say,
test1.mp3 and test2.mp3.  Then when I connect enough clients to test1.mp3,
the streamTranscoder instance producing that stream usually starts
misbehaving seriously after a few minutes - but test2.mp3 with fewer or no
listeners is still running fine!  The test1.mp3 doesn't recover until the
sT instance producing it is restarted: once the problem has appeared,
it doesn't help to eg. disconnect the clients from the stream.

I'm genuinely puzzled by this.  How can the sT instance possibly care
about its output stream being loaded with listeners at the Icecast server?
:-/
> (note: if the Shoutcast setup uses the same bandwith and users, then
> this is not the problem)
Yes, it does.  And I've failed to reproduce this bug with Shoutcast so
far.  Will keep trying, though.

-- 
Antti N?yh?                 "Ich bin der Zorn Gottes;
http://www.sairwas.org/         wer sonst ist mit mir?"

Antti Nayha wrote:
> I'm genuinely puzzled by this.  How can the sT instance possibly care
> about its output stream being loaded with listeners at the Icecast server?
> :-/
It doesn't, at least not directly.  That is rather odd.  You *definitely* 
have enough CPU?  Transcodes can be expensive.  Also, have you looked in 
the Icecast error log?  I would expect the problem to be mroe likely in 
Icecast than ST, but who knows?

Geoff.