[Halld-offline] Data Challenge Meeting Minutes, October 22, 2012

Mark M. Ito marki at jlab.org
Mon Oct 22 23:00:15 EDT 2012


Find the minutes below and at 

   -- Mark

GlueX Data Challenge Meeting, October 22, 2012

      * CMU: Paul Mattione, Curtis Meyer
      * JLab: Eugene Chudakov, Mark Ito (chair), David Lawrence, Elliott
      * UConn: Richard Jones


      * Mark has [16]tagged a release of sim-recon for use in the first
        data challenge. The plan is to also use the previously tagged
      * Elliott announced that he and David are working on
        raw-data-from-simulation. Many pieces are there, but it does not
        look close enougfh to include in this next data challenge.

Minutes from last time

    We reviewed the [17]minutes from the October 8th meeting. We noted that
    Curtis wrote to IT division requesting a 50% share of the farm, and the
    request has been approved, at least in principle. Right now, about a
    third of the processing power of the farm resided in new nodes running
    CentOS6 and are lightly used and so we effectively have a good chunk of
    what we requested.

Setting conditions for DC-1

    Mark proposed making a directory in the Subversion repository to
    document conditions for the software. It would have a text document
    calling out the versions to be used and the actual configuration files
    to be read by the various programs. He will propose a location.

    Richard will construct a gridmake file that will determine the
    configuration to be run on the grid. He requested a simple wiki page
    with the relevant information, a sensible idea.

    We discussed some details and came to some decisions.
      * We will simulate events from the coherent peak between 8.4 and 9.0
        GeV only.
      * We will not apply any level 1 trigger algorithm to the "raw data"
        but David will modify the REST format to encode the result of the
        level 1 plug-in. This must be done during the reconstruction stage
        since the result depends on the raw hits.
      * Paul will work on a plug-in to make simple monitoring histograms
        during reconstruction so we have a handle on the results of each
        output file without further analysis. Mark had planned to use the
        output of the phystree plug-in, but that is likely more data than
        we need for this purpose.

Site status

      * JLab Mark has started a mini-challenge on the JLab farm using the
        tagged version and the new farm nodes.
      * CMU Paul has built the new tagged version. He will run some jobs
        with it.
      * Grid Richard will wait for the final decision on versions and then
        create the corresponding gridmake package.

    Retrieved from



More information about the Halld-offline mailing list