[Halld-offline] Software Meeting Minutes, August 4, 2020

Mark Ito marki at jlab.org
Tue Aug 4 19:24:17 EDT 2020


Please find the minutes here 
and below.

   --  Mark

    GlueX Software Meeting, August 4, 2020, Minutes

Present: Alex Austregesilo, Thomas Britton, Mark Ito (chair), Richard 
Jones, Naomi Jarvis, David Lawrence, Susan Schadmand, Beni Zihlmann

There is a recording of his meeting 
<https://bluejeans.com/s/NwcD67rD2@l/> on the BlueJeans site. Use your 
JLab credentials to authenticate.


Draft of DSelector documentation 
See Beni for the link to edit the Overleaf document.

      Review of Minutes from the Last Software Meeting

We went over the minutes from July 21 

        Corrupt CCDB SQLite Files

Mark reported that instances of corrupt CCDB SQLite files have occurred 
several times over the past two weeks. Recall that the error is "Lost 
connection to MySQL server." The new size requirement has been catching 
them and bad ones are not getting shipped to Oasis.

        Compiler upgrade discussion

Mark reported that Richard's HOWTO 
on installing and running cvmfsexec <https://github.com/cvmfs/cvmfsexec> 
allowed him to easily install Oasis on Mark's RHEL7 box at the Lab. 
This, coupled with our standard container, allows almost instant access 
to a JLab-like development/running environment, as advertised. With a 
CentOS 8 container, which is in the works, it could ease our transition 
to a more advanced version of GCC.

      Restoration of Execution Tests for Pull Request Builds

is working now. Sean Dobbs may have more to say at the next meeting. 
Mark pointed out that there is new environment set-up scheme to ensure 
consistency between building and running the binary tests.

      Sluggish Response on halldweb.jlab.org

Several people have been noticing periods of slow response from our main 
webserver, halldweb.jlab.org, including, not not limited to, use of the 
wikis. Yesterday morning the server was timing out on requests, not good 
at all. Mark reported that during slow-downs, the webserver has plenty 
of idle CPU cycles and does not appear to be swapping. It is the case 
that the majority of web requests during these times are from the 
MCwrapper Dashboard, at the level of a few Hertz from multiple browser 
clients. Thomas assured us that those operations are light weight and 
cannot account for the slow-downs. Going forward:

  * Thomas has increased the period between updates requests from the
    browser application, despite his assertion that those requests
    cannot possibly be the problem, at Mark's request.
  * Naomi suggested that people submit ServiceNow requests (write an
    email to helpdesk at jlab.org) whenever a problem is encountered. That
    might raise the visibility of the issue.
  * Mark mentioned the possibility of putting up a dedicated server,
    either a webserver, database server, or both, to move the load away
    from other essential functions on halldweb.
  * Mark also suggested that the Computer Center implement some sort of
    history mechanism that might help identify the bottleneck.

      dE/dx theta Correction

Naomi reported that her improvements to the CDC dE/dx measurement 
located on this branch 
are ready to go. She expressed her concern that the dE/dx quantities 
would undergo a sudden change if this branch were merged, making those 
quantities inconsistent with those encoded in REST files from previous 
reconstruction launches. Mark said that similar improvements are merged 
all the time. Beni gave a strong suggestion that the pull request be 
composed. [Added in press: Naomi submitted the pull request and Beni 
merged it with the comment "too good to not have."]

      Python 3 Compatible Build System

Mark described the changes, announced last week 
that allows us to build our software on either a Python-2-based system 
or on one based on Python 3. See his slides 
for details. This work is a first step at developing a container system 
for CentOS 8.

      Review of recent issues and pull requests

David called our attention to halld_recon Issue #418, *hd_root hangs at 
the end of evio file with is_valid_run_end = false*, originally 
submitted by Naomi. Richard will have a look.

      Review of recent discussion on the GlueX Software Help List

We went over two items:

  * JLab logbook image upload problem
    <https://groups.google.com/g/gluex-software/c/tDLG5qcStjA>. We heard
    that Mark Dalton has received enough feedback to proceed.
  * Simulation stuck at first event
    <https://groups.google.com/g/gluex-software/c/FFKZCBWIAgI>. Igal
    Jaegle is still having this problem. Richard will take a look.

Retrieved from 

  * This page was last modified on 4 August 2020, at 19:21.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-offline/attachments/20200804/cbb490e1/attachment-0001.html>

More information about the Halld-offline mailing list