I'm having problem with an Airtime installation. After a while the system runs out of memory and kills the "airtime-liquids" process.
I'have another server running in the same webradio, it's a twin system but with less music loaded, and it does not have this problem. And on another system wich also has 2GB of RAM and about 1TB of music, no problem too.
Attached here there are the recent Airtime logs, the syslog portion of the killing episode and the little snippet of liquidsoap code that I replaced in ls_script.liq.
no, that server should not rebroadcast streams! In fact, there are no "Web Streams" defined! :? It is still almost always running on "automatic mode" with the ls_script.liq patch with a direct stream to a local server... where can I check for this rebroadcast in the logs?
Maybe I did some misconfiguration in the ls_script.liq... misteriosamente, Francesco P
The only changes I did are the ones in the snippet that I attached to the first message! Just to be sure, I did a diff with a newly downloaded tarball...
sorry to be late! Yes, the two machines are twins, both in hardware and in software. The only thing in which they differ is the amount of music loaded on them: about 24G on the failing one, only 9GB on the one that is working fine.
Maybe it is so. On a third machine, with 2GB of RAM and 1TB of music, there is no "out of memory" problem, but sometime I get a segfault from airtime-liquidsoap...
However, I'm reverting from playlist() to request() in that background-music-script, I'll see if things start going better! :)
It would be really good if you reported those problems. The liquidsoap team is really good at fixing bugs.
On Tue, Dec 4, 2012 at 6:03 PM, FrancescoP <<br />airtime-support@lists.sourcefabric.org> wrote:
> Maybe it is so. On a third machine, with 2GB of RAM and 1TB of music, > there is no "out of memory" problem, but sometime I get a segfault from > airtime-liquidsoap... > > However, I'm reverting from playlist() to request() in that > background-music-script, I'll see if things start going better! :) > > *speratamente, > Francesco P.* > >
Just for the record, I think that the problem I was slowly investigating has been already found and fixed in the meanwhile (they got the same conclusion that I got in the meanwhile):