GlueX Software Meeting, March 17, 2020
GlueX Software Meeting
Tuesday, March 17, 2020
3:30 pm EDT
Remote meeting, please use BlueJeans: 968 592 007
- 1 Agenda
- 2 Minutes
- 2.1 Announcements
- 2.2 Review of Minutes from the Last Software Meeting
- 2.3 Review of Minutes from the Last HDGeant4 Meeting
- 2.4 Corrupt EVIO Files
- 2.5 Review of recent issues and pull requests
- 2.6 Review of recent discussion on the GlueX Software Help List
- 2.7 Action Item Review
- Review of Minutes from the Last Software Meeting (all)
- Review of Minutes from the Last HDGeant4 Meeting (all)
- Corrupt EVIO Files (Beni)
- Review of recent issues and pull requests:
- Review of recent discussion on the GlueX Software Help List (all)
- Action Item Review (all)
Present: Alex Austregesilo, Thomas Britton, Sean Dobbs, Mark Ito (chair), Naomi Jarvis, David Lawrence, Keigo Mizutani, Zisis Papandreou, and others perhaps[?]
There is a recording of his meeting on the BlueJeans site. Use your JLab credentials to access it.
- Compute Canada Award Zisis announced the award of 265 core-years on Compute Canada. See the link for details.
- Thomas asked if there was a use-it-or-lose-it aspect to the award with respect to the amount of awards in subsequent years. If so he could arrange it so that Compute Canada would be favored for GlueX simulation jobs. Zisis will get back to Thomas on this.
- New Version Set released: version_4.16.0.xml This version set came out last week. It includes the new version of CCDB (see the next announcement). Also, this file the does not appear in the DIST directory.
- Note that the DIST directory has not disappeared. All files that have been served from there in the past remain there with the exception of the version set files, which are now in HALLD_VERSIONS.
- New sub-minor release: CCDB 1.06.07. This version has substantially improved performance on the MySQL servers.
Review of Minutes from the Last Software Meeting
We went over the minutes from March 3.
Recon Launch halld_recon vs. Analysis Launch halld_recon
We had further discussion of the recon-launch/analysis-launch halld_recon version issue.
- Thomas pointed out that the problem is complex. The fact that the analysis library is part of halld_recon is the root cause. There is no easy way out of the situation.
- He also mentioned that the pick-an-analysis-launch approach could be done, but it does not solve all problems. In particular, there are users who want simulated REST files and have no interest or knowledge about which analysis launch they should align with. In addition, changes to MCwrapper-bot to implement this are not simple. He also reminded us that the current system is not missing functionality in the sense that you can do nearly everything you would reasonably want to do with it as it stands.
- Sean suggested documentation so that users with well-defined use cases can figure out what choices to make on the bot-gui.
- Alex reminded us that the recon launch versions themselves need get rebuilt from time to time with patches necessary to make them compatible with new halld_sim versions, another complication.
- Alex suggested trimming down the number of choices on the bot-gui. That could be done but requires an expert.
- Mark admitted that there are some recon-launch compatible versions sets that have yet to be built against version set 4.16.0 (mentioned above).
CCDB Access Control
Mark spoke to Dmitry Romanov about this since the last meeting. The access control is available in CCDB 2.0 and can be implemented once that becomes the standard. Dmitry said that a full-blown administrative interface to configure the access control scheme is not there; that is going to take a lot of Python coding. On the other hand the scheme in its current state is configurable; configuration is an expert operation probably requiring use of the MySQL command line. So we will accept that and use the feature.
Review of Minutes from the Last HDGeant4 Meeting
We went over the minutes from March 10 without significant discussion.
Corrupt EVIO Files
Beni reported jobs crashing on recently taken raw data files. The problem has been traced to a single channel of a Flash-250 in ROC 17. That module has been replaced. Richard Jones has observed the problem in the data already taken and is instituting a fix so that programs can skip over the problematic blocks of data. David expressed interest in reviewing the solution once it becomes available.
Review of recent issues and pull requests
- Alex will revisit halld_recon Issue #116 "Vertex z dependent reconstruction efficiency" to see if it is still with us.
- Sean asked people to review changes in halld_recon Pull Request #332 "Timing monitoring updates" to make sure the guidance is correct and the expert email lists are as they should be.
Review of recent discussion on the GlueX Software Help List
We went over the list.
Mark reported that the CCDB problem Igal Jaegle reported over the weekend was due to a full root partition on hallddb.jlab.org, our main MySQL server. We were not able to determine the cause of the problem, but Marty Wise has added disk space to the VM; the partition was running at 36% full when he last checked.
Action Item Review
- Trim choices in MCwrapper-bot. (?)
- Write a wiki page to describe how to navigate MCwrapper-bot. (Thomas)
- Implement a pick-your-analysis-launch option to MCwrapper-bot (Thomas, long-term)
- Build recon-launch compatible version sets with the latest versions of halld_sim. (Mark)
- Review changes to monitoring plot guidance from Sean. (all)