That does sound a bit too high (should be more around 20 MB for the parameters 
that you mention).  Memory utilization in 0.7.0c should not be higher than 
0.7.0b (rather lower).  These days I use valgrind with the command from the 
attached shell script to profile gnunetd's memory utilization (note that 
valgrind itself increases the footprint quite a bit), and I've yet been 
unable to detect any problems with 0.7.0c.  The expected values are 
documented in doc/memoryuse.txt (in svn).

So if you want to help, it would be great if you could use valgrind/massif to 
gather some profiles -- that would help me see where the problem might be.  
Note however that with valgrind your CPU usage will go up significantly, 
which in turn may lower the #connections gnunetd will establish (and create 
possibly other artefacts that one would not observe without valgrind). 

And I guess I might mention the obvious workaround: restart gnunetd (not a 
good solution, but if you need a solution *now*, that's your best bet ;-)).  
Otherwise, please report any findings (as in, valgrind traces, etc.) to 
Mantis so that we can try to address the issue.

Cheers!

Christian

On Tuesday 11 April 2006 07:31, David Kuehling wrote:
> Hi,
>
> did something about gnunetd's memory consumption change from from
> version B to C?  top shows currently 116Mb of resident memory which is
> much more than I had previosly seen.
>
> How is that memory consumption related to datastore size and number of
> connected peers?  According to gnunet-stats, my datastore is currently
> approaching the 5GB mark and 17 peers are connected.
>
> regards,
>
> David
valgrind --tool=massif --alloc-fn=xmalloc_unchecked_ --alloc-fn=xmalloc_ 
--depth=10 --alloc-fn=xgrow_ bin/gnunetd -d -L DEBUG
_______________________________________________
Help-gnunet mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gnunet

Reply via email to