GNOME Bugzilla – Bug 335423
Evolution Leaking Memory
Last modified: 2006-06-09 02:36:38 UTC
Please describe the problem: I wanted to create a bug report that Evolution seems to be leaking. After 2 hours of use: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 22985 drichard 16 0 219m 96m 20m S 0.0 0.6 0:53.86 evolution-2.6 and then after re-launch: 11215 drichard 16 0 100m 24m 16m S 0.0 0.2 0:01.49 evolution-2.6 Items performed in the 2 hours: Send and receiving email Send and accept meetings drop and drag files into new email messages. With 220 concurrent users, any leaks that you can plug are appreciated. Steps to reproduce: 1. 2. 3. Actual results: Expected results: Does this happen every time? Other information:
Created attachment 62345 [details] [review] Patch to plug some leaks in evolution I found the same problem, after two weeks of continuous use (by only one user), evolution was using over 500 MB! Valgrind revealed quite a few leaks in evolution, evolution-data-server, and gtkhtml, but also orbit and gtk are leaky too. The attached patch fixes some of the more obvious leaks in evolution. I'm sure there are more.
Created attachment 62346 [details] [review] Patch for evolution-data-server Another memory leak patch, this time for evolution-data-server. Should this be moved to a separate bug report?
Created attachment 62347 [details] [review] Patch for gtkhtml A memory leak patch for gtkhtml. I should mention that these are pretty well tested. I've been using them for a few days now with no crashes.
harish/varadhan: Can these patches be reviewed and put into the pre-2.6.1 build for testing? I will gladly put down any memory leak patches and QA them.
harish, varadhan, any news on this?
I'm watching the servers run with 100s of people on it and watching for leaks. It seems to be running OK in this regard. I'll update this report with additional information.
I had reviewed and committed the gtkhtml patch from the e-p list to the HEAD which unfortunately did not have a bug reference. Updating the patch status.
Chris, you may commit these patches to HEAD and the gnome-2.14 branch. If you are unable to do so, pl. ping me back. I shall do the commits. Thanks
Harish: Applying to 2.6 stable as well?
yes. gnome-2.14 branch hosts Evolution 2.6.x.
Adding a "me-too". Left evolution running over the weekend. Evolution processes are using over 500 Megs of RAM. Using Evolution 2.6.1 with outlook connector. Gnome: 2.12.3 This could be causing clock-applet to leak memory too: http://bugzilla.gnome.org/show_bug.cgi?id=343912
2.6.2 RPMS are starting to appear and should start to contain some of these fixes. I'm watching our new 2.6.2 build right now to see how it's running in terms of memory.
hey chris, again thanks very much for your patches, great stuff that is always welcome! i have now committed the e-d-s patch both to cvs head and the stable branch, so the e-d-s patch weill be included in the unstable evolution 2.7.3 and the stable evolution 2.6.3. feel free to commit the evolution patch yourself if you have cvs write access. thanks a lot for your patience.
i have now also committed the evolution patch both to cvs head and the stable branch, so it will be included in the unstable evolution 2.7.3 and the stable evolution 2.6.3. /addressbook/gui/merging/eab-contact-merging.c (cancelit) had been already fixed and committed to head and stable branch to fix bug 336713. feel free to provide further bug reports and/or patches, we really appreciate it.