GlueX Offline Meeting, September 28, 2016
GlueX Offline Software Meeting
Wednesday, September 28, 2016
2:00 pm EDT
JLab: CEBAF Center F326/327
- 1 Agenda
- 2 Communication Information
- 3 Minutes
- Recent Fedora and Ubuntu releases are too advanced, need to use ROOT 5
- After collab meeting: run-dependent geometry proposal from Sean
- Group disk from 700 to 800 GB
- sim-recon on Mac OS X
- Review of minutes from the last meeting (all)
- Collaboration Meeting Agenda
- SciComp Meeting Report (Mark)
- Launches (Paul M./Alex A.)
- Software Review
- Review of recent pull requests (all)
- Review of recent discussion on the Gluex Software Help List.
- Action Item Review
- The BlueJeans meeting number is 968 592 007 .
- Join the Meeting via BlueJeans
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2016 on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2016/ .
- CMU: Naomi Jarvis, Curtis Meyer, Mike Staib
- JLab: Alexander Austregesilo, Alex Barnes, Thomas Britton, Brad Cannon, Paul Eugenio, Mark Ito (chair), David Lawrence, Paul Mattione, Nathan Sparks, Simon Taylor, Beni Zihlmann
There is a recording of this meeting on the BlueJeans site.
- We are in strange situation where recent Fedora and Ubuntu releases use GCC compilers that are too advanced for ROOT6. ROOT5 must be used. This situation is not expected last too long; there are fixes on the ROOT Git master branch.
- Sean Dobbs has a proposal for managing run-period-dependent shifts in detector geometry. He will describe it at the next Offline Meeting (after the Collaboration Meeting).
- Our group disk has been expanded from 700 to 800 GB.
- To build sim-recon on Mac OS X you need the C++ compiler from Xcode 8. You need to be on El Capitan or higher.
Review of minutes from the last meeting
- We looked at minutes from the meeting on September 14. We mainly discussed the upcoming Software Review (see below).
Software Review Preparations
- The first all-Hall meeting on review prep was held last Monday.
- We took a quick look at GlueX Note 3113, being produced to support our presentations (linked from the Software Review 4 wiki page). Curtis has contacted identified contributors and described the content we would like them to provide. The document is being version controlled at GitHub.
- At the last Hall D meeting prep meeting, we made initial assignments of topics to talks.
Collaboration Meeting Agenda
We reviewed the current state of the offline software session on the agenda for Thursday:
- 16:20 Session IIb (100) --- Offline/Calibration
- 16:20 (20) --- Overview --- Mark Ito
- 16:40 (30) --- Calibration --- Sean Dobbs
- 17:10 (20) --- Offline Data Processing --- Alexander Austregesilo
- 17:30 (30) --- Data Analysis Studies --- Paul Mattione
- 17:40 (20) ---
There is room for another talk. Please contact Mark if you have something you would like to present. Thanks to Sean for putting this together.
SciComp Meeting Report
Mark went over his notes from the meeting on September 22.
- There is still a plan to institute a formal process for requesting time on the batch farm.
- New farm nodes are to installed soon and will be running CentOS7. The native compiler is GCC 4.8.5, good enough for our software. Although a CentOS7 ifarm node is available now (ifarm1402), it does not have all of the packages we need to build our software yet.
- The Lustre problems we have been experiencing have been seen on similar systems at Fermilab.
- Machines from 2011 are about to be retired from the farm. Beni asked if it was possible for us to acquire some and deploy them in the Counting House. Mark will look into it.
The ver02 reconstruction launch had problems with less-than-optimal calibration constants. Although the analysis launch from that launch ran, Paul M. and Alex A. decided to run another analysis on ver01 REST files, using updated plug-ins. This Analysis Launch ver04. All four batches are complete now.
David asked if the analysis launch was I/O bound. The answer is not known at present.
Review of recent pull requests
We went over the list.
There is currently a problem producing compressed HDDM files introduced at Pull Request #544. The b1pi test has been failing as a result. Richard Jones is working on the fix.
Beni raised the problem of maintaining all of the SConstruct files in all of the private plugin directories scattered around the Collaboration. Recently several folks could not build there plug-ins due to a change in the standard SConstruct. No easy solution occurred to us. We will have to discuss this in the future.
Review of recent discussion on the GlueX Software Help List
We looked at the menu of recent topics.
Due to  recent security changes at JLab, web-based programs may need access to a JLab CA certificate to work properly. In particular, GitHub access using HTTPS may be affected. Sean posted a solution on the GlueX Offline FAQ.
- Look into getting old batch farm nodes for the Hall D use.
- Create a solution to the SConstruct in Private Plugins Maintenance Problem.
- Contribute your sections to the Software Review Document.
- Fix compression in the HDDM libraries.