Dieter, It's OK, I figured it out ... It's odd, once the leak stopped, the performance got worst. But it was partially due to me. I managed to make a variety of tweaks and changes that solved the problem, and it's running faster than ever now! While I've got your ear ... Is it possible to prefetch objects in the threads' object cache? So that instead of loading objects on demand, I could force the loading at start-up? I have some well know sets of objects that I'd like to have this done to, helping the startup time ... And possibly also allowing better indication of memeory usage. (It's a high traffic website with agressive caching, it can be slow after startup for a while, I'd like to see if that can improved). Thanks AGAIN, I owe you :) J.F. -----Original Message----- From: Dieter Maurer [mailto:dieter@handshake.de] Sent: May 27, 2004 1:53 PM To: Jean-Francois.Doyon@CCRS.NRCan.gc.ca Cc: zope@zope.org Subject: Re: [Zope] Help! Serious performance problem! (PTS? Zeo? SMP?) Jean-Francois.Doyon@CCRS.NRCan.gc.ca wrote at 2004-5-27 10:28 -0400:
For reasons I can't seem to grasp, my Zope has gone real slow on me ...
I switched from using Localizer + TranslationService to using the collective's PlacelessTranslationService (PTS) 1.0fork-rc7 ... And since then this occurs.
How about using my "ZopeProfiler" and determining where the time is spent? <http://www.dieter.handshake.de/pyprojects/zope> -- Dieter