After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 439780 - Evolution-data-server uses up WAY too much memory
Evolution-data-server uses up WAY too much memory
Status: RESOLVED INCOMPLETE
Product: evolution-data-server
Classification: Platform
Component: general
1.8.x (obsolete)
Other All
: Normal critical
: ---
Assigned To: Evolution Shell Maintainers Team
Evolution QA team
Depends on:
Blocks:
 
 
Reported: 2007-05-19 21:43 UTC by Joel Schaerer
Modified: 2007-06-27 09:27 UTC
See Also:
GNOME target: ---
GNOME version: 2.15/2.16


Attachments
Massif backtrace (3.29 KB, text/plain)
2007-05-24 16:23 UTC, Joel Schaerer
Details
Massif backtrace (31.45 KB, text/plain)
2007-05-24 16:32 UTC, Joel Schaerer
Details
Massif backtrace (58.34 KB, application/postscript)
2007-05-24 16:32 UTC, Joel Schaerer
Details

Description Joel Schaerer 2007-05-19 21:43:43 UTC
Please describe the problem:
Here's a copy of the script I've just added to my /etc/cron.hourly:

#!/bin/bash
killall -v -r ^evolution-data-serv.* 2>&1 | logger -t joel

Yes, killing EDS every hour makes my whole system a lot more useable! I have two rather sizeable IMAP accounts. However, this is no reason for evolution-data-server to use up to 800Mo of memory on my 1GB system, causing the OS to start swapping and the whole system to slow down to a crawl.

I don't know the precise cause of the problem, I just know that I've had it ever since I've been running evolution. This might not be a very useful bug report, it IS a serious issue and should be tracked and acted upon.

Steps to reproduce:
1. Start evolution
2. Wait for a few hours
3. Watch system collapse


Actual results:
evolution data server uses too much memory.

Expected results:
A reasonable amount of memory (say, 100MB) is used

Does this happen every time?
Yes

Other information:
Comment 1 Joel Schaerer 2007-05-19 22:53:24 UTC
Maybe I should add I'm using Fedora Core 6.
Comment 2 Ross Burton 2007-05-24 07:25:50 UTC
If you could run e-d-s inside massif we'd be able to see where your EDS is allocating huge amounts of memory from.  Something like this will make massif useful:

GLIB="--alloc-fn=g_malloc --alloc-fn=g_realloc --alloc-fn=g_try_malloc --alloc-f
n=g_malloc0 --alloc-fn=g_mem_chunk_alloc --alloc-fn=g_strdup --alloc-fn=slab_all
ocator_alloc_chunk --alloc-fn=g_slice_alloc --alloc-fn=g_slice_alloc0 --alloc-fn
=e_memchunk_alloc --alloc-fn=e_memchunk_alloc0 --alloc-fn=g_hash_node_new --allo
c-fn=g_hash_table_insert"

DB="--alloc-fn=__os_malloc_eds --alloc-fn=__os_realloc_eds --alloc-fn=__db_r_att
ach_eds

valgrind --tool=massif --depth=5 --alloc-fn=posix_memalign --alloc-fn=memalign -
-alloc-fn=valloc $GLIB $DB evolution-data-server-1.10
Comment 3 Joel Schaerer 2007-05-24 16:23:49 UTC
Created attachment 88741 [details]
Massif backtrace
Comment 4 Joel Schaerer 2007-05-24 16:25:49 UTC
I ran EDS with massif with the options you gave an attached the backtrace. EDS didn't run for long, maybe a couple of minutes, then it seemed to terminate and relaunch itself.

It did seem to leak some memory when I opened and closed Gaim repeatedly, maybe the problem could come from there? I do use the adress book synchronization feature between gaim contacts and evolution.
Comment 5 Joel Schaerer 2007-05-24 16:32:26 UTC
Created attachment 88742 [details]
Massif backtrace
Comment 6 Joel Schaerer 2007-05-24 16:32:47 UTC
Created attachment 88743 [details]
Massif backtrace
Comment 7 Ross Burton 2007-05-24 16:49:22 UTC
Did you use evolution as you normally do with massif running?  This didn't really leak or use that much memory really.
Comment 8 Joel Schaerer 2007-05-24 16:52:36 UTC
As I said, it only ran for a few minutes before it terminated itself, it usually takes a few hours before it really starts using huge amounts of memory. I'll try to run it again if I have the time and see if it stays up longer.
Comment 9 Ross Burton 2007-05-24 16:57:38 UTC
e-d-s terminated in your run because nothing was using it.  Run it in massif and when it has started just start Evolution and view the addressbook, that will stop it quiting.
Comment 10 Joel Schaerer 2007-05-24 16:58:35 UTC
I just did a little more testing (outside of massif, i'll try again with massif when I have more time). EDS leaks about 5Mo of memory each time I restart gaim. I'm not sure if that's what entirely caused my problem, though.
Comment 11 Joel Schaerer 2007-05-24 16:59:14 UTC
Just saw your last comment. I will try again tomorrow. Thanks for the help!
Comment 12 André Klapper 2007-06-14 14:47:06 UTC
joel, any news here?
Comment 13 Joel Schaerer 2007-06-14 14:59:09 UTC
I tried once for a few hours, but unfortunately my computer crashed before I had the chance to retrieve the output. Unfortunately I haven't had the time to rerun it since then.
Comment 14 Joel Schaerer 2007-06-27 09:24:16 UTC
Ok, I have dropped using evolution so I won't be able to provide the necessary information on this bug. You may close it.