OWG Meeting 4-Dec-2013

Video Conferencing

We will be connected via ESnet using the standard GlueX number: 8542553

Agenda

*** Turn on conference equipment ***

• Announcements
  • Network down because switch unplugged!
• Review of minutes from 20-Nov-2013 meeting
• CODA3 Release - Graham, Dave A
• Policies for Using Online Directories and Accounts - Dave L, Mark D, Elliott
• Terminal servers and ROC setup - Bryan, Hovanes, Mark D
  • boot messages
  • BIOS access
  • remote mods vs bringing ROC upstairs
  • ordering cables and connectors, custom connectors for XVB601 controller consoles
  • ts_connect, kernel drivers and where to install, etc.
• Spring 2014 Online Data Challenge - Mark D, Dave L, Elliott
• What should we use as the standard GCC compiler for the online cluster? - Hovanes, Dave L, Elliott
  • single default? setup script? something else?
• Farm Manager - Vardan, Dave L, Elliott
• DAQ testing status - Mark D, Dave A
• Trigger and FCAL testing status - Alex S, John L, Mark D
• Telecom fiber installation status - Elliott
• Trigger fiber installation schedule - Chris C
• RootSpy - Sean
• Online databases - Elliott, Dmitry
  • what level of unification is possible?
  • CCDB, conDB, trigDB, translation table, SiPM, EPICS/Nerses, etc.
• JInventory database update - Elliott, Serguei P
  • FCAL, BCAL/SiPM, FDC, etc.
• EPICS update - Hovanes, Vanik
  • Alarm system, HV/LV, etc.
• Controls update - Dave B, Elliott
  • FCAL darkroom, BCAL, Gas system, CDC, PS, solenoid, etc.
• DAQ group status reports - Bryan M, Dave A, Graham, Carl, Vardan
  • GIT, EVIO, cMsg, CODA3, ECS/CSS, disentangler, readout list, Java/cMsg GUI
• Electronics and trigger status reports - Chris, Alex, Fernando, Ben, Ed, William, Bryan M
• Mantis Task Tracker
• Midas Elog

• Upcoming topics
  • raw data, compressed vs uncompressed
  • Calibration/Alignment issues?
  • Status of processing uncompressed and compressed simulated raw data and monitoring/L3 farm - Dave L, Elliott
  • Ed to talk more on FPGA programming

Time/Location

1:30 PM Wed 4-Dec-2013 CC F326

Next Meeting

1:30 PM Wed 18-Dec-2013 CC F326

Minutes

Present: Elliott W, Bryan M, Curtis M, Dave L, Mark I, Graham H, Elton S, Sean D, Justin S, Yi Q, Eugene C, Simon T, Hovanes E, Serguei P, Mark D, Beni Z, Dave A, Vanik K.

CODA 3.0

Dave A said that a release should be installed by the end of the week, although a few patches might be expected for this developing software system. This led to a discussion of when to patch a release and when to create a new release, with a major concern being the stability of releases for others using it. The discussion ended without a clear strategy other than to see how things go. This first release will be make by hand by Dave from the DAQ group svn repository, he will eventually automate the procedure. Other questions were how to back out of an update to a release, will the different halls get different patches, compiler version conflicts, are global svn tags needed in the DAQ group svn repository and should branches be used. Most of these were not resolved.

Online Policies

Elliott presented and GlueX adopted the policies in the link above. These relate to how accounts are used and what should be done in them, who should have sudo privileges, ability to edit DNS entries, etc.

Terminal Servers

Bryan and Dave strongly advised using terminal servers to connect to the ROC serial port. Hovanes previously ordered about 9 servers that Elliott and Serguei will install in the hall. Serial connections can be used to view console output, boot messages, to edit the BIOS (Bryan has to enable this). We will try to use the tsconnect package successfully used in CLAS, which requires installation of a special kernel driver on selected computers. Elliott and/or Serguei will order required parts and connectors, Paul L will be asked to install the driver.

Spring Online Data Challenge

Consensus is to hold an extensive data challenge in May with many milestones leading up to it. The milestones and test plan needs to be consistent with the 12GEV schedule. The main goal is to check the full data path from ROC to silo, including the L1 trigger system and the event monitoring system.

Compilers

Many months ago the Online Group approved use of the new C++11 standard, which required a relatively modern compiler (GCC 4.8.0). RHEL6, operating system on all our computers, was delivered many years ago with an older compiler that was not C++11 compliant (GCC 4.4.7). This led to a discussion as to which should be the default compiler, the old one in /usr/bin or newer ones in /gapps (/apps on the CUE machines). The online code management and release system transparently supports multiple compilers, so this might not be a big concern other than it appears that in a mixed system the main program must be linked with the more recent compiler and the LD_LIBRARY_PATH must be set correctly (not entirely sure of this). Eugene said we should just stick with the old compiler and not waste time on compiler incompatibilities. Doing this means all code developed using C++11 features must be rewritten.

Farm Manager

Vardan is working on this, should be ready soon. Will need to be tested with JANA monitoring and L3 processes.

DAQ Testing

Mark D reported on many bug fixes, creation of COOL configurations, some problems with mutex locking and the TI. He has successfully read out 6 crates with block level greater than one. No testing of multi-stage event building yet.

RootSpy

Sean gave a status update, see his slides above. He has a few more weeks of work left to complete his upgrade plans.

Trigger and FCAL Test

Installation in progress, including temporary pulsing system for bases.

EPICS

Work progressing on FDC LV system, one mainframe so far. No power yet, need more work on HV setup. Vanik has been working on microscope motors and drivers, and 3-axis synchronized motion. We have received embedded computers for the microscope system, work on low-level communication drivers in progress (to match UConn firmware conventions). Hovanes is testing the EPICS alarm system BEAST, he has build the MySQL database and is in the process of testing this facility. An important question is how to generate alarms from non-EPICS channels. Note the the alarm server uses JMS for publish/subscribe messaging via a public domain JMS implementation.