On Sun, Nov 04, 2012 at 05:43:05PM -0500, Martin Konečný wrote: > One thing to note is that the nice level of the replaygain analysis is set > so that that it has low CPU scheduling priority, while the Liquidsoap > process is given high scheduling priority. > > This replaygain analysis is done ahead of time, and once all tracks in the > library have been analysed, it should never become a problem again. From > the command line we can run > > sudo -u postgres psql -c 'SELECT count(*) FROM cc_files WHERE replay_gain > IS NULL;' airtime > > If the print value is zero, this means that no more replaygain analysis is > happening (unless more track are added of course) and therefore is not the > problem. > > At the moment all signs still point to Apache CPU usage spike as the > culprit.
Yes, that makes sense. Replaygain does not appear to be running anymore. The Apache CPU usage makes a lot of sense, since the problem didn't happen when nobody was on the system, but seemed to escalate as more users were using the web interface, and as one user kept the interface open at all times on his laptop. The system is running only airtime, no other sites or processes. I have to ask, does the airtime-liquidsoap process run with RT priority on a system that has an Ingo RT kernel? The system will probably be split up into two different machines in the springtime, or, may be replaced with better hardware. Thanks again for all your help on this. -ken ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_nov _______________________________________________ Savonet-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/savonet-users
