[Halld-offline] High memory usage in hd_ana

[David Lawrence]

Hi Will,

   I have been able to reproduce this problem using revision 8610 and 
your file. However, the problem does not seem to present itself with the 
most recent version of the code (8643). It looks like there were some 
changes to the PID and tracking libraries in revisions 8623 and 8621 
that might have resolved the problem (perhaps without meaning to).

   When running the revision 8610 code, I did see that the problem would 
not show up due to a single event. I could narrow it down to a range of 
about 118 events in your file where it needed the first event in the 
range to be analyzed, but didn't start blowing up the memory until the 
last event in the range. Other events in between may have contributed. 
It was clear that this was not a slow memory leak, but one that occurred 
suddenly as you described.

   Can you try again with tagged release sim-recon-2011-12-13 and see if 
the problem is gone for you as well?

Regards,
-David

On 12/14/11 8:46 PM, Will Levine wrote:
> I'm running into a problem with this hddm file:
> http://www-meg.phys.cmu.edu/~wilevine/memory_issue/memory_issue.hddm
>
> When I run:
>
> hd_ana memory_issue.hddm -PPLUGINS=danahddm
>
> memory usage stays relatively flat for a while then suddenly jumps to
> several GB, at which point the program stops analyzing events and must
> be killed. The problem always arises in the same part of the file
> (around event 3509), but I was unable to isolate a single event that
> caused the problem.
>
> Program is compiled from svn revision 8610 with no modifications.
>
> Richard can chime in if this sounds similar to the problem that he is seeing.
>
> Will
> _______________________________________________
> Halld-offline mailing list
> Halld-offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-offline