[Orca-users] Re: memory leak

[Liston Bias]

3 systems sounds very low unless you have lots of devices that you are
trying to map.  I used to run 100 systems on 4Gig Memory system and I had
to start orca multiple times a day.  I've since upgraded to 16Gig and took
out many of the device mapping I don't care about and usage went down a
lot, but still restart daily.

Hopefully, Blair or someone else on the list can shed light on your
pariticular issue.  Let us know how resolved.

- Liston

On Fri, 30 May 2003, Berezin, Dmitry wrote:

> Well, one or two hundred megs would be OK, but I'm talking about
> gigabytes!!!
> Here is the snapshot of orca running for just about 2 days:
>
> ps -o user,pid,ppid,vsz,stime,tty,time,etime,args -p 5120
>
>  USER   PID  PPID     VSZ   STIME TT     TIME     ELAPSED COMMAND
>  orca  5120     1 3594784  May_28 ?  05:17:40  1-23:51:00
> /usr/local/bin/perl -w /opt/orca/bin/orca -d /opt/orca/lib/orcallator.cfg
>
> It has already grabbed over 3GB of memory.
> ...and it only graphs 3 systems in this installation.
> Again, I did not have this problem with the previous version.
>
>   Blair?
>
> Thank you,
>
>   -Dmitry.
>
> >> I too seem to be running into a memory leak with Orca on Solaris 8.  Orca
> >> was holding steady yesterday for a while at 9MB, but this morning it is
> >> using 136MB!  Using Orca/Orcallator, I can tell that Orca dramatically
> >> stepped up its memory usage twice, about three and nine hours after it
> >> started.  Other than those jumps, its memory usage appears to be stable.
> >
> > Orca/perl does a poor job of releasing memory usage.  If you run orca in
> > daemon mode, I suggest you try restarting it automatically at regular
> > individuals to clear memory usage.  I have a cron that does this daily.  I
> > have not had any problems with graphs (about 300 systems) doing this, but
> > other may have.
>
>
>
>