[Halld-offline] Notes on jobs for data challenge

[Sean Dobbs]

Hi all,

I've been running sets of test jobs on our cluster over the weekend to make
sure that we're ready for the data challenge.  I've started keeping notes
here:

https://halldweb1.jlab.org/wiki/index.php/NU_DC2_Tests

The short version of the story so far is that with jobs of 10K events each,
I'm getting a >50% success rate, though I don't have the hard memory limits
that seemed to be causing some of the crashes that Mark mentioned on
Friday.  The failed jobs mainly die at REST creation, and while I've found
increasing the JANA thread timeout limit to be helpful, the problems seem
consistent with either certain events taking too long to process, or some
site-specific bottlenecks.


Cheers,
Sean


-- 
Dr. Sean Dobbs
Department of Physics & Astronomy
Northwestern University
phone: 847-467-2826
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://mailman.jlab.org/pipermail/halld-offline/attachments/20140209/e236c435/attachment.html