On Mon, May 14, 2012 at 2:23 PM, kevin dawson <ke...@dawson10.plus.com> wrote: > from #Padre http://irclog.perlgeek.de/padre/2012-05-14#i_5587625 > > 04:43 szabgab good morning > 04:43 Finally I managed to track down the reason why the server was > spiralling in extreme load lately > 04:44 or rather the immediate cause was something in our trac > installation > 04:44 for nowI disabled it > 04:45 I'll try to fix it soon but now at least all the other site > might be responsive > > do we know when Padre's trac will be in operation again, > > how long is this outage for? > > what are these web sites that take priority over padre-trac?
I have to say that this sentence made me really angry. Is this some kind of not-too-well hidden accusation? But let me list just a few of the sites on the same server that was hardly available in the last few days: - The SVN of Padre - The main Padre web site - The Perl Community Adserver which is embedded in a number of other web site make all those other web sites seem partially broken. - the web site of the Perl Weekly - The archives of this (and all the other mailings lists I run) - CPAN::Forum - perl.org.il and a few more. And now for the situation: The sever was hitting 10-15 load and because historically I saw the CPAN::Forum causing trouble and as I was in Ukraine at the time I kept restarting the server for few days. Something was strange though. Even after killing all the instances of Apache and letting the server calm down to 0 load, one I started Apache I saw the load immediately. Today in the morning I remove all the VirtualHost configurations and added the back one-by-one trying to see which one causes the immediate hike in the load. It turned out that it is the padre.perlide.org VirtualHost. Insdide that, after commenting out the trac/ part and starting Apache, the load remained low. That's where I am now. There are two directions I can follow for fixing this: 1) Try fixing this server 2) Move to a new server with the latest of everything I'll try to look for 1) but I have also been planning 2) anyway so I might just try to do that soon. Anyone has any idea what could have caused the problem? Was something added to trac about a week ago that might be different than what we had there earlier? I searched "trac causing high load" but I don't think that any of the hits there help me. regards Gabor _______________________________________________ Padre-dev mailing list Padre-dev@perlide.org http://mail.perlide.org/mailman/listinfo/padre-dev
