OWG Meeting 9-Oct-2013

From GlueXWiki
Jump to: navigation, search

Video Conferencing

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


Agenda

*** Turn on conference equipment ***

  • Announcements
  • Review of minutes from 11-Sep-2013 meeting
  • ROC order - Dave A
  • Trigger/DAQ requirements for FCAL tests - John L
  • Dec 2013 Online Data Challenge - Elliott
  • ROC and crate IP naming scheme - Hovanes
  • Crate data rates and networking strategy - Mark D, Elliott
  • Moving controls computers to 26 subnet - Elliott
  • CODA3 installation and DAQ testing - Mark D, Dave A
  • Farm Manager - Vardan, Dave L, Elliott
  • JInventory database update - Serguei P.
  • EPICS update - Hovanes, Nerses, Vanik
  • Controls update - Dave B, Elliott
    • Solenoid, FCAL darkroom, BCAL, Gas system, CDC, etc.
  • DAQ group status reports - Bryan M, Dave A, Graham, Carl, Vardan
    • Farm manager, 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
      • Linux SBC configuration, installation, booting - by Bryan
      • Infiniband installation and management - by Dave A
    • Ed to talk more on FPGA programming


Time/Location

1:30 PM Wed 9-Oct-2013 CC F326


Announcements

Next Meeting

1:30 PM Wed 23-Oct-2013 CC F326


Minutes

Present: Elliott W, John L, Mark D, Bryan M, Serguei P, William G, Hovanes E, Sean D, Matt S, Justin S, Carl T, Vardan G, Simon T, Dave L, Eugene C, Curtis M, Dave A, Alex S, Beni Z.


ROC Order

Dave reported that the final order (40 I think) is on hold due to budget uncertainties. He's ready to proceed when it is allowed.


FCAL Testing

  • Crates and cables installed.
  • One month needed to seal dark room to allow HV to be turned on.
  • Power immanent (end of Oct), if it doesn't come soon we implement an interim power plan for selected crates. Full crate needs 30A/208V.
  • Will start testing with LED pulser.
  • Need paddles for cosmic ray testing, investigating where to get them from. Also will need HV for paddles.
  • Start with single crate readout, move to multi-crate using master TI boards (max 9 crates). There are 12 crates total in two rack areas.


Dec 2013 Online Data Challenge

Two modes are planned:

  1. read triggered data from front-end boards (random pulser, i.e. just pedestals) from whatever crates have power and are connected to the trigger system. "Event splicer" downstream of final EB will add missing ROC data to event so that monitoring/L3 farm gets complete events.
  2. inject MC data into data stream in all ROC's no front-end board readout.


Crate Naming Scheme

Serguei will implement pretty much what Hovanes presented a long time ago. E.g. DAQ crate FDC1 will have IP name rocfdc1, its fan tray will get IP name vmefdc1, etc. CODA uses the logical name FDC1 to find the IP name. Some devices have multiple NICs so Serguei will come up with a naming scheme. Finally there is still some confusion about how to use the DAQ vs controls networks.


Crate Data Rates

Mark and Elliott worked out a fiber installation strategy based on crate data rates that Dave determined via running the EVNT plugin from the previous data challenge on more data. See the link above for complete details.

  • Dave A noted that practical network and backplane rates can be far below ideal rates.
  • On the FADC125 the current bottleneck is inter-chip communication, not backplane or network communication. (later...Ed says this likely can be eliminated with modified firmware).
  • Peak rates can be higher than maximum sustained rates since everything gets smoothed out eventually.
  • Need to measure real rates with actual triggered beam data, of course.
  • Is our max beam rate 10**8 or 5 10**7 photons/sec? The latter is used in our PAC proposal.
  • In any case we're planning for 10**8.
  • Not clear how to split crates if this is ever needed...it will be a real pain!


Moving Controls to Hall Subnet

Dave B will move all PLC stuff to the 26 subnet (controls) from the lab-wide 36 subnet. gluon01 will remain on the 36 subnet because it is used by other setups around the lab. Brent will allow MySQL access to gluon01 from the 26 subnet so that we maintain connection to the PLC database on gluon01.


CODA3

  • Version currently installed and up to date.
  • Dave A said he wants to reinstall it differently for general use.
  • Global trigger crate control still to be integrated.


Farm Manager

Vardan working diligently, should have working version by the end of the month. Need to agree on failure modes and their consequences. He can test it standalone prior to getting it running in the hall. First version will be simple w/o fancy features, we'll add to it as needed.


JInventory

Installed for production use. Now everyone just has to be diligent about keeping it up to date! Still need requirements from Fernando concerning maintenance component.


EPICS

  • Nerses and Vanik working on HV/LV for BCAL.
  • Hovanes working on alarm system (BEAST from ORNL/SNS). We likely need cMsg input to the alarm system.
  • Student from Santa Maria coming soon to work on BCAL controls.
  • Accelerator needs to get working on applications they agreed to do a long time ago!


Controls

George and Elliott meet Tues mornings individually with folks working on updating the solenoid control system (i.e. no more weekly group meetings since the solenoid is sort of on the back burner now).


DAQ

Bryan says CENTOS6 will be available on the ROCs soon, currently using V5. Some concerns about partitions using RHEL5 vs RHEL6.