[icecast] test load for icecast2 ?

[Michael Smith]

On Wednesday 05 March 2003 21:14, Jörn Nettingsmeier wrote:
> On Wed, 5 Mar 2003, Michael Smith wrote:
> > On Wednesday 05 March 2003 19:13, Joern Nettingsmeier wrote:
> > > i'm test-driving an icecast2 relay on a rather crappy network
> > > connection. to hammer on it, i was listening to the stream with xmms
> > > while doing a couple of parallel wget -O - http://..mystream 2&>1
> > >
> > >  >/dev/null from another machine. however, as few as 5 parallel wgets
> > >
> > > would reproducably stall the streaming .
> > > is wget not a suitable test load, or did i do something wrong in my
> > > icecast setup ?
> >
> > icecast has no way to prioritise its network traffic over what other
> > things are doing - so if you have lots of other traffic taking up as much
> > bandwidth as they can (wget gets stuff as fast as possible), it'll tend
> > to drown out the (close-enough to constant for these purposes) bandwidth
> > usage of icecast2.
>
> i'm aware of that, but i used wget on the *stream*, and that should not
> come in any faster than with the stream's bitrate, right ?
> i was wondering if it's wget that's confusing the icecast2 server, or if
> my setup just isn't up to the load...

Oops. My apologies - I misread what you originally said. Just how crappy _is_ 
your "rather crappy network connection"? wget should work fine.

>
> how do you guys do stress-tests ?

Usually "nc". Though I ran out of cpu last time I did that at around 3000 
clients (on an oldish machine, and older icecast2 code).

Mike

--- >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.