Nikos Kouremenos wrote:

On 3/11/06, Jonas Ådahl <[EMAIL PROTECTED]> wrote:
I'm running the latest (atleast not more than a couple of days old) SVN
code, and I've noticed some really bad memory issues [1]. As you can see
in the graph something is happening between 10h and 15h and somewhere at
32h. I didn't do anything with the client at 32h. I suspect that it
doesn't have anything to do with what I do with the client, since I was
sleeping from ~28h to 35h :P.

hoh that's interesting unless you sleap walk (or I should say sleap

can you try svn up, which remove psyco which results in Gajim using
less memory? (and hopefully using it better) and tell us if the
results remain.

btw, how did you keep those stats and how did you produce from them the graph?

thanks a lot
I did a fresh checkout, will monitor memory usage again for some of days then :)

The stats were generated fairly simple. I just ran this:
while [ 1 ] ; do ps -eo pid,rss|grep 25204|awk '{ print $2 ; } ' >> GAJIMlogg ; sleep 1h; done where 25204 is the PID of the python process running and "rss" is (taken from man ps) "resident set size, the non-swapped physical memory that atask has used (in kiloBytes)".

This will create a file with the number of kiloBytes the process is using on each line. Then I use this file, doing some regexp magic, and then creating a matlab script that plots the data:
where y is the 1-dim-matrix containing the values from the ps output.


Attachment: signature.asc
Description: OpenPGP digital signature

Gajim-devel mailing list

Reply via email to