I just successfully reproduced this with an Ogg stream as well. I killed the Ogg mountpoint within Icecast and saw this in netstat:
tcp4 0 0 192.168.1.99.55772 X.X.X.X.8000 CLOSE_WAIT tcp4 0 0 192.168.1.99.65159 X.X.X.X.8000 ESTABLISHED I have liquidsoap encoding an AAC+ and Ogg stream. Interestingly enough, the AAC+ stream stayed connected (which I would expect, given that I only killed the Ogg stream), but it's buffer quickly drained down to 0% and required me to restart liquidsoap. Suggesting that liquidsoap was stuck? This is a bit of an odd situation. I have had no problem using liquidsoap on FreeBSD in the past. Unfortunately this computer is at a client's location that I don't have physical access to. I suspect that they may have a firewall or router that could be causing the connection to not shut down properly? I just tried using the Ices source encoder as well. It also did not reconnect right away either, but only took about 2 minutes compared to the 6 minute timeout that I've been seeing in liquidsoap. Here is the ices log: [2010-05-11 10:13:34] INFO om_shout/check_shout_connected Connected to server: X.X.X.X:8000/mount-32k.ogg [2010-05-11 10:13:34] DBUG om_shout/output_ogg_shout initialising output stream [2010-05-11 10:16:41] EROR om_shout/output_ogg_shout Failed to write to X.X.X.X:8000/mount-32k.ogg (Socket is busy) [2010-05-11 10:16:41] DBUG om_shout/_output_connection_close closed shout connection [2010-05-11 10:16:44] DBUG om_shout/check_shout_connected Time we started stream on X.X.X.X:8000/mount-32k.ogg Is there any type of connection timeout variable in liquidsoap that I should look at using? Andrew On 2010-05-10, at 1:52 PM, David Baelde wrote: > Hi, > > This could be a valid bug report, but we have to figure out if it's a > general problem in our icecast client code, or if it's specific to > external encoders and in particular AAC (which I suspect). The use of > an external binary for encoding, whose output is then sent to icecast, > implies some normal complications and also some really hairy problems > (the behavior of the aac encoder may change slightly from one version > / platform to the other, etc). > > Can you reproduce this problem with an MP3 or OGG icecast output? > > If the problem is AAC specific, you should try using the SVN version > which has a quite different code regarding those aspects. We won't fix > 0.9.2 for such problems (it's not a stable release, only a "snapshot") > but we'll try to get the SVN and upcoming 1.0 as good as possible > (it'll be hard to be perfect for AAC, I'm afraid). > > Cheers, > -- > David ------------------------------------------------------------------------------ _______________________________________________ Savonet-users mailing list Savonet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/savonet-users
