<html>
  <head>

    <meta http-equiv="content-type" content="text/html; charset=utf-8">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Folks,<br>
    <br>
    Please find the minutes below and at<br>
    <br>
     
<a class="moz-txt-link-freetext" href="https://halldweb.jlab.org/wiki/index.php/GlueX_Offline_Meeting,_July_8,_2015#Minutes">https://halldweb.jlab.org/wiki/index.php/GlueX_Offline_Meeting,_July_8,_2015#Minutes</a><br>
    <br>
      -- Mark<br>
    ____________________________________<br>
    <br>
    <div id="globalWrapper">
      <div id="column-content"><span dir="auto">Minutes, GlueX Offline
          Meeting, July 8, 2015</span><span class="mw-headline"
          id="Minutes"></span><br>
        <div id="content" class="mw-body" role="main">
          <div id="bodyContent" class="mw-body-content">
            <div id="mw-content-text" dir="ltr" class="mw-content-ltr"
              lang="en">
              <br>
              Present:
              <br>
              <ul>
                <li> <b>FIU:</b> Mahmoud Kamel<br>
                </li>
                <li><b>JLab</b>: Alex Barnes, Mark Dalton, Mark Ito
                  (chair), David Lawrence, Paul Mattione, Curtis Meyer,
                  Eric Pooser, Matt Shepherd, Justin Stevens, Simon
                  Taylor, Beni Zihlmann</li>
                <li> <b>NU</b>: Sean Dobbs</li>
              </ul>
              <br>
              <span class="mw-headline" id="Announcements">Announcements</span><br>
              <ol>
                <li> <b>Sim-Recon 1.3.0.</b> Mark reminded us of the <a
                    rel="nofollow" class="external text"
href="https://mailman.jlab.org/pipermail/halld-offline/2015-June/002076.html">latest
                    release</a>. It is compatible with JANA 0.7.3 and
                  uses a new BCAL smearing scheme from Tegan Beattie.
                  <ul>
                    <li> CLHEP is no longer needed by AmpTools and comes
                      packaged with Geant4 in recent versions. It will
                      be dropped from the packages built by default by
                      the GlueX Version Management System (GVMS). </li>
                  </ul>
                </li>
                <li> <b>Legacy build directories.</b> Some old
                  directories in the /group/halld/Software/builds
                  directory at JLab are about to disappear. Lately they
                  have been a source of confusion. See <a
                    rel="nofollow" class="external text"
href="https://mailman.jlab.org/pipermail/halld-offline/2015-July/002080.html">Mark's
                    email message</a> for details.</li>
              </ol>
              <br>
              <span class="mw-headline"
                id="Review_of_minutes_from_June_24">Review of minutes
                from June 24</span><br>
              <br>
              We went over the <a
href="https://halldweb.jlab.org/wiki/index.php/GlueX_Offline_Meeting,_June_24,_2015#Minutes"
                title="GlueX Offline Meeting, June 24, 2015">minutes
                from last time</a>.
              <br>
              <ul>
                <li> The switch to the new volatile disk hardware went
                  without a hitch.</li>
                <li> The Data Center power outage went well as well.</li>
                <li> Mark I. is still promising to send out instructions
                  on how to switch from use of one of the various builds
                  of GlueX software to another.</li>
                <li> Mark and Beni have succeeded in building the
                  complete suite of GlueX software on Beni's Ubuntu
                  machine. The next step is to set up a nightly build.</li>
                <li> Beni has succeeded in creating a working build of
                  GlueX software against a version of ROOT 6. We will
                  need to discuss an upgrade in the future.</li>
              </ul>
              <br>
              <span class="mw-headline"
                id="Offline_Monitoring_Train_Schedule_and_Status">Offline
                Monitoring Train Schedule and Status</span><br>
              <br>
              Paul reported that the plan is to launch the next offline
              monitoring train this Friday. There was no train last
              week.
              <br>
              <br>
              Sean gave us an update on offline monitoring
              post-processing. <a rel="nofollow" class="external text"
href="https://halldweb.jlab.org/wiki/images/2/27/Sdobbs_OfflineMtg_OffMon_July2015.pdf">His
                slide</a> reads as follows:
              <br>
              Offline Monitoring Backend
              <br>
              <ul>
                <li> The system is ready for the next run </li>
                <li> Memory usage under control - ~1 GB per process </li>
                <li> pylint is your friend </li>
                <li> Parallelized hadd - phadd
                  <ul>
                    <li> Uses python thread pooling </li>
                    <li> Current settings: 6 threads, 10 files in a
                      batch. On ifarm, limited by bandwidth to work disk
                    </li>
                    <li> Run 2439: ~40 min (hadd) -> ~14 min (phadd)
                    </li>
                    <li> N.B. hadd -n processes batches serially, uses
                      less memory but slower </li>
                  </ul>
                </li>
                <li> Ready to monitor simulations </li>
                <li> Processing of online histograms works now, too</li>
              </ul>
              <br>
              He also mentioned:
              <br>
              <ul>
                <li> The next step is a webpage to show results.</li>
                <li> We need a similar set of procedures for simulation.</li>
                <li> We should think about running something for the
                  online as well.</li>
              </ul>
              <br>
              <span class="mw-headline"
                id="Spring_2015_Commissioning_Simulations">Spring 2015
                Commissioning Simulations</span><br>
              <br>
              Sean updated the status of the <a
href="https://halldweb.jlab.org/wiki/index.php/Spring_2015_Commissioning_Simulations"
                title="Spring 2015 Commissioning Simulations">simulation</a>
              effort.
              <br>
              <br>
              We have to finalize the monitoring plugins we want to run
              and then we will be ready to go. Test jobs have run
              successfully.
              <br>
              <br>
              Justin explained why some jobs are timing out. We moved
              the boundary between using Pythia and Eugene's PDG
              compilation of decay modes down to 2.5 GeV from the
              nominal 3.0 GeV, an energy below the coherent peak.
              Otherwise we would get no omega's or phi's at the peak
              energy. This causes bggen to go into an infinite loop
              sometimes. There is still an acceptable fraction of jobs
              which complete normally so we are just accepting that
              yield.
              <br>
              <br>
              <span class="mw-headline" id="Proposal_on_Git_for_GlueX">Proposal
                on Git for GlueX</span><br>
              <br>
              Mark ran down a proposal for the change from Subversion to
              Git. See <a rel="nofollow" class="external text"
                href="https://halldweb.jlab.org/talks/2015/git_for_gluex_3.pdf">his
                slides</a> for the details. One of the slides addressed
              the procedure for switch-over:
              <br>
              <ol>
                <li> Comment: Only HDDS and sim-recon will be converted.</li>
                <li> Comment: Having two authoritative repositories is a
                  contradiction in terms.</li>
                <li> Schedule switch-over for COB, Wed., July 15.</li>
                <li> Between now and then, people should check in
                  changes to the Subversion repository.</li>
                <li> Continue to get experience using practice
                  repositories.</li>
                <li> Do final conversion from Subversion to Git
                  overnight, Wed. to Thu. Practice repositories will be
                  deleted.</li>
                <li> Open for business for Git after all-clear
                  announcement Thursday morning.</li>
              </ol>
              <br>
              We had a vote and there were no votes opposing the change.
              <br>
              <br>
              <span class="mw-headline" id="JANA_0.7.3">JANA 0.7.3</span><br>
              <br>
              David described changes in the new release of JANA. See <a
href="https://halldweb.jlab.org/wiki/images/4/41/20150708_JANA0.7.3.pdf"
                class="internal" title="20150708 JANA0.7.3.pdf">his
                slides</a> for details. He highlighted a few of the new
              features:
              <br>
              <ul>
                <li> <b>janaview plugin</b>. A system for click and
                  point browsing of event objects.</li>
                <li> <b>support stdin types in janaroot</b>. Automatic
                  creation of ROOT trees for selected factory-made
                  objects.</li>
                <li> <b>Fixed longstanding bug in jgeomread where
                    attributes specified in the middle of the xpath were
                    not displayed</b>.</li>
                <li> <b>JObject:Get() does recursive searches for
                    associated objects</b>.</li>
                <li> <b>Event Barriers</b>. Ability to synch event
                  processing threads on special events such as EPICS
                  events.</li>
                <li> <b>Added --listconfig switch</b>. Dump the JANA
                  configuration to the screen</li>
              </ul>
              <br>
              <span class="mw-headline" id="Geant4_Update">Geant4 Update</span><br>
              <br>
              Richard gave a report.
              <br>
              <ul>
                <li> There is a project on GitHub now: <a
                    rel="nofollow" class="external free"
                    href="https://github.com/rjones30/HDGeant4">https://github.com/rjones30/HDGeant4</a>
                  .</li>
                <li> He described the <a rel="nofollow" class="external
                    text"
                    href="http://nvie.com/posts/a-successful-git-branching-model">workflow</a>
                  he is supporting for development work.</li>
                <li> Building the code using the magnetic field
                  mechanism used in DANA has been solved.</li>
                <li> David has suggested starting with Geant4 10 since
                  that incorporates multi-threaded particle tracking.
                  Because of custom fixes to the Geant4 code that
                  Richard has developed, the same code will not run
                  under both version 9 and 10.</li>
                <li> The single particle gun works now among other
                  features.</li>
              </ul>
              <br>
              <span class="mw-headline" id="multi-threaded_mcsmearing">multi-threaded
                mcsmearing</span><br>
              <br>
              David described a study he did that compared performance
              of mcsmear with different settings of output compression
              and different numbers of threads. See <a rel="nofollow"
                class="external text"
                href="https://logbooks.jlab.org/entry/3344045">his log
                book entry</a>. Adding threads has very little effect on
              execution speed, but requesting compression adds over a
              factor of two to the execution speed. In the usual context
              where we are interested in reconstructing the data as soon
              as it is smeared, the intermediate smeared file will not
              be shipped around and compression helps very little.
              Richard pointed out that the connection between mcsmear
              and DANA could be through a named pipe to avoid the disk
              write/read overhead altogether.
              <br>
              <br>
              <span class="mw-headline" id="Mixing_Simulated_Events">Mixing
                Simulated Events</span><br>
              <br>
              We decided to defer Sean's talk until next time.
              <br>
              <br>
            </div>
            <div class="printfooter">
              Retrieved from "<a dir="ltr"
href="https://halldweb.jlab.org/wiki/index.php?title=GlueX_Offline_Meeting,_July_8,_2015&oldid=68547">https://halldweb.jlab.org/wiki/index.php?title=GlueX_Offline_Meeting,_July_8,_2015</a>"</div>
          </div>
        </div>
      </div>
      <div id="footer" role="contentinfo">
        <p><br>
          <br>
        </p>
      </div>
    </div>
    <br>
  </body>
</html>