How large should a mogilefsd process become?

[komtanoo.pinpimai at livetext.com]

I did a memory leak test for the cvs version, attached in this email.
Replicator and worker did really leak in the test. However, I've found a
work around by restarting these processes internally at the point where
they are safe to restart and it looks good.

-kem

On Fri, August 11, 2006 2:36 pm, Mark Smith wrote:
>> Thanks for the hint (and to the other posters suggesting the same root
>> cause) but I can say definitely that we have only small objects. The
>> largest object is 5MB, the second largest 1.2 MB. All objects together
>> add up to only 1.3 GB.
>
> Very strange...
>
>
> Well, the replication process is getting a review/rework currently, so
> we'll keep an eye out for anything that would cause it to bloat.
>
>> Before I go and edit the source to add GTop statements: are there any
>> debugging options that I can try first?
>
> You can telnet in to the port your tracker listens on and type '!watch'
> which will reveal errors/status updates from the various jobs (including
> the replicators).
>
> You can also turn on debug output and run the tracker in the foreground,
> but that's pretty spammy.  To do this, run the tracker and set the
> environment variable DEBUG to 1 or 2.  (2 is more output!)
>
>
> --
> Mark Smith
> junior at danga.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.danga.com/pipermail/mogilefs/attachments/20060812/f1d56f23/memleak-0001.htm