Hi All, A few weeks ago I posted this message to the list but we still haven't resolved our memory leak, mainly because we have no idea where to look. So having trawled around a few forums I decided to install the LeakFinder product onto our development server and click around a little watching the refcounts. Here I noticed that Products.Silva.Image.Image & DateTime.DateTime.DateTime had the largest number of references. So following the LeakFinder instructions I attempted to patch each of these classes. I started with the Products.Silva.Image.Image class and continued to click around to return some results, unfortunately this never returned any results at all so that wasn't much help. Next I patched the DateTime.DateTime.DateTime class which returned a huge amount of information on the results tab. So I'm now hoping someone could help me in deciphering these results and explain what exactly should I be looking for here? And it would also be useful to know why the Products.Sivla.Image.Image class doesn't return any results at all. Hope someone can help. Jon -----Original Message----- From: zope-bounces@zope.org [mailto:zope-bounces@zope.org] On Behalf Of Jonathan Bowlas Sent: 15 February 2006 11:49 To: zope@zope.org Subject: RE: [Zope] Memory Leak Debugging Hi All, We're having serious problems with performance running Silva CMS on Zope 2.7.8-final, python 2.3.5, linux2 set up on a machine with 2.5GB RAM. There are currently only about 65 users and around 2000 documents in the CMS but it seems to run for a week, then the RAM fills up and crashes so we have to restart the server manually. The Zope logs aren't much help as I enquired on the Zope list and their response was we need more RAM. But surely 2.5GB is enough. I've also checked Control_Panel/DebugInfo for any refcounts with significant jumps and Products.Silva.Image.Image seems to be the culprit: 11.32am 11.36am Products.Silva.Image.Image 56203 58153 +1950 Acquisition.ImplicitAcquirerWrapper 44825 45151 +326 So we've tried to figure out which kind of request triggers these refcount jumps by creating a separate instance with a very small ZODB Cache (so it gets filled up soon), and clicked around a little, watching the refcounts but we cannot seem to replicate it. So basically this is a plea for help from an expert Zope debugger who can assist us in finding the source of the problem so we can resolve it. Or alternatively someone who can give us a set of instructions that we can follow that will enable us to locate the problem, I've googled it and there are no clear answers. FYI we have the following products installed on our Zope instance. Annotations 0.4.3 BTreeFolder2-1.0.1 DocFinderTab (0.5.2) External Method-1-0-0 FileSystemSite 1.4.2 Formulator 1.10 Groups 0.4 IssueTrackerProduct (0.6.13) LDAPUserFolder (2.3) MIMETools MailHost-1-3-0 OFSP-1-0-0 PageTemplates-1-4-0 ParsedXML 1.4 PlacelessTranslationService (1.0) PluginIndexes ProxyIndex 1.2 PythonScripts-2-0-0 Sessions Silva 1.4.1 SilvaBlog 0.4.2 SilvaCustomAutoTOC 0.9 SilvaDocument 1.4.1 SilvaDocumentPatches 1.3b6-1 SilvaExtETHLDAP 0.7 SilvaExternalSources 0.10.3 SilvaMetadata 0.10 SilvaNewsNetwork 2.0 SilvaReferenceChecker 0.2 SilvaStylesheets 0.6.2 SilvaUCLAttributes 0.7 - (A UCL product*) Silva UCL Gallery 0.1 - (A UCL product*) SilvaUCLSubjectAudienceFilters 0.6 - (A UCL product*) SilvaViews 0.11 SiteAccess-2-0-0 SiteErrorLog Sprout 0.6.2 StandardCacheManagers-1-1-0 TemporaryFolder Transience Tropos (1.9.12) XMLWidgets-Pluggable 0.13 (CVS/UNRELEASED) ZCTextIndex ZCatalog-2-2-0 ZGadflyDA ZMySQLDA 2.0.8 ZODBMountPoint ZOracleDA ZReST (1.1) ZSQLMethods Zope Tutorial 1.2 kupu 1.3.2 z3locales 0.1 If you require any further information do not hesitate to ask. Hope someone can help. Jon Bowlas University College London _______________________________________________ Zope maillist - Zope@zope.org http://mail.zope.org/mailman/listinfo/zope ** No cross posts or HTML encoding! ** (Related lists - http://mail.zope.org/mailman/listinfo/zope-announce http://mail.zope.org/mailman/listinfo/zope-dev )