[Halld-offline] Software Meeting Minutes, September 15, 2020

[Mark Ito]

Folks,

Please find the minutes here 
<https://halldweb.jlab.org/wiki/index.php/GlueX_Software_Meeting,_September_15,_2020#Minutes> 
and below.

   -- Mark

__________________________________


    GlueX Software Meeting, September 15, 2020, Minutes

Present: Shankar Adhikari, Alex Austregesilo, Thomas Britton, Mark 
Dalton, Sean Dobbs, Mark Ito (chair), Igal Jaegle, Naomi Jarvis, Simon 
Taylor, Nilanga Wickramaarachchi, Beni Zihlmann

There is a recording of his meeting 
<https://bluejeans.com/s/bDwJNbkUkhV/> on the BlueJeans site. Use your 
JLab credentials to authenticate.


      Review of Minutes from the Last Software Meeting

We went over the minutes from September 1. 
<https://halldweb.jlab.org/wiki/index.php/GlueX_Software_Meeting,_September_1,_2020#Minutes> 


  * The space situation with the work disk is better now. There is about
    10% (or 20 TB) free now
    <https://halldweb.jlab.org/disk_management/work_disk_2.png>. Thanks
    to those who helped in the clean up.
      o The work disk leader board
        <https://halldweb.jlab.org/disk_management/work_report.html> in
        back in operation. Go there to see who is using the most space
        on the work disk.
  * Alex's problem with a large number of NIS lookups in DSelector jobs
    has resolved itself. We are not sure why.


      Talk at OSG All-Hands Meeting

Mark I. reviewed some of the points he made during his talk 
<https://halldweb.jlab.org/doc-public/DocDB/ShowDocument?docid=4669> at 
the Open Science Grid All-Hands meeting 
<https://indico.fnal.gov/event/22127/> a little over a week ago. The 
talk was meant to emphasize the relationship between the EIC and the OSG.

  * Computing and software management each have at least two co-leaders,
    one of whom is from BNL and one of whom is from JLab.
  * Computing hardware resources are also envisioned to be contributed
    from both of the labs.
  * GlueX figures prominently in the list of OSG users from medium-sized
    collaborations.
  * CLAS12 used about 5 million core hours during the month of July,
    comparable to GlueX usage for the entire year.


      PID matching suggestions based on data/MC comparisons

Sean led us through the first slide (see below) of his talk 
<https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=4652> on 
future areas for data vs. Monte Carlo comparisons. He will give a 
complete presentation at next week's HDGeant4 Meeting.

Sdobbs pid 20200908.png 
<https://halldweb.jlab.org/wiki/index.php/File:Sdobbs_pid_20200908.png>

 From the discussion:

  * On widening TOF cuts as a function of momentum:
      o Simon reported that there is already a hook to do this in the
        PIDFOM system.
      o Alex reported that there is also a mechanism available for a
        momentum-dependent cut when doing particle identification in the
        Analysis Library.


      Review of recent issues and pull requests

Thomas brought up a recent pull request to the halld_sim repository 
<https://github.com/JeffersonLab/halld_sim/pull/153> from Alex. This 
fixes a problem where simulation jobs can go into an infinite loop if 
the CCDB does not show any flux for a corresponding real data run. The 
change was made to the beam properties class.


      New guidance for selecting version sets with MCwrapper

Sean described his new wiki page, How to choose software versions on the 
MC submission form 
<https://halldweb.jlab.org/wiki/index.php/How_to_choose_software_versions_on_the_MC_submission_form>. 
This led to discussion on related topics.

  * Shankar asked for advice on how to specify the RCDB query to be used
    for a range of runs. Turns out that each run period has a different
    query and they are not called out on the form. This is because the
    "best" underlying query to the RCDB is different for each run
    period, thus different query aliases are needed for each instance.
    This needs better documentation.


      Specifying Simulation Runs

Mark D. noted that when filling out the MCwrapper web form, it is hard 
to remember the "correct" responses, those that worked well in a 
previous project. Thomas pointed out a feature ("mod=2") where 
parameters from a previous project can be copied into a new form, with 
the exception of some fields that probably should not be re-used, such 
as the output file location.


      Need for a New Version Set

Sean and Shankar pointed out that we are due for a new version set that 
incorporates recent improvements to several packages. Mark I. agreed to 
produce one.


      Action Item Review

 1. Ask the Computer Center about staging the start of jobs in time.
 2. Make sure that the automatic tests of HDGeant4 pull requests have
    been fully implemented.
 3. Create a new version set with the latest improvements. (Mark I.)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20200924/38ed8d63/attachment.html>