Difference between revisions of "March 9, 2016 Calibration"

From GlueXWiki
Jump to: navigation, search
(Agenda)
(Minutes)
Line 55: Line 55:
  
 
== Minutes ==
 
== Minutes ==
 +
 +
Attending: Sean (NU); Simon, Mark D., Paul M., Will, Justin, Nathan, Adesh, Mark I., Eugene (JLab); Matt S. (IU); Curtis, Naomi (CMU); Mahmoud (FIU); Cristiano (MIT)
 +
 +
* Announcements
 +
** There was a discussion of farm usage based on [https://mailman.jlab.org/pipermail/halld-offline/2016-March/002276.html this email] from Mark Ito yesterday.
 +
*** Over the weekend, we "stress tested" the farm with calibration and monitoring/reconstruction jobs.  This led us to be using more than our "share" of farm resources for an extended period of time (~50% compared to a share of 30%).  A large portion of the farm had been converted to "exclusive" nodes based on our expected usage, and there was talk of converting nodes back.
 +
*** We had previously let SciComp know that we expected our usage this spring to be high, based on the data we are collecting.  Mark I. is planning to discuss raising our allocated share of farm resources.
 +
*** Paul reported a scaling factor of 35 on the exclusive nodes.  Each has 24 physics cores + 24 hyperthreads, so that corresponds to each hyperthread being ~45% of a physical core for these jobs.  Sean will work on extracting useful numbers for calibration jobs.
 +
*** There was talk of adding a "12 core" job queue based on poorly understood utilization numbers.  No one knows the current number of machines that are configured to be in which queue.  This should be on the SciComp website.  Mark I. also pointed out that the usage numbers in the email he forwarded should be available to everyone.  This could help us monitor our own usage.
 +
** Sean asked if anyone had looked at the mode 8 data that we've taken yet (planned as ~5-10% of the total production data).  This data was to be used to check our reconstruction.  No one had, though there is some confusion between which runs are mode 8 production data, and which are used for trigger studies (which often have detectors taken out of the readout).
 +
** Mark I. pointed out that on the [https://halldweb.jlab.org/wiki/index.php/GlueX_Offline_FAQ#SWIF offline FAQ], there is a SWIF section, which contains some useful information (including undocumented commands!).  He encouraged people (especially Paul and Sean) to add to this.
 +
** Mark I. has tagged [https://github.com/JeffersonLab/sim-recon/releases/tag/1.10.0 a new version of sim-recon]
 +
* Calibration Train
 +
**
 +
 +
 +
Larrieu

Revision as of 16:04, 9 March 2016

GlueX Calibration Meeting
Wednesday, March 9, 2016
11:00 am, EST
JLab: CEBAF Center, F326

Communication Information

Remote Connection

You can connect using BlueJeans using the meeting number 630 804 895 .       (Click "Expand" to the right for more details -->):

  1. Make sure you have created a BlueJeans account via your JLab CUE account using this link:

  2. Meeting ID: 630804895
    • (you may need to type this in, depending how you connect)

  3. If connecting via Web Browser: click this link (no passcode is needed):

  4. If connecting via iOS or Android App:
    • Use your JLab e-mail address to log in and then enter the meeting ID given above to join the meeting

  5. If connecting via Phone: Dial one of the following numbers and then enter the meeting ID above and hit "#" or "##"

  6. If connecting via Polycom unit:
    • Dial 199.48.152.152 or bjn.vc
    • Enter meeting ID above
    • Use *4 to unmute

Slides

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/ .

Agenda

  1. Announcements
  2. Calibration Tasks
  3. Subdetector Reports
  4. Sim1 progress
  5. AOB

Minutes

Attending: Sean (NU); Simon, Mark D., Paul M., Will, Justin, Nathan, Adesh, Mark I., Eugene (JLab); Matt S. (IU); Curtis, Naomi (CMU); Mahmoud (FIU); Cristiano (MIT)

  • Announcements
    • There was a discussion of farm usage based on this email from Mark Ito yesterday.
      • Over the weekend, we "stress tested" the farm with calibration and monitoring/reconstruction jobs. This led us to be using more than our "share" of farm resources for an extended period of time (~50% compared to a share of 30%). A large portion of the farm had been converted to "exclusive" nodes based on our expected usage, and there was talk of converting nodes back.
      • We had previously let SciComp know that we expected our usage this spring to be high, based on the data we are collecting. Mark I. is planning to discuss raising our allocated share of farm resources.
      • Paul reported a scaling factor of 35 on the exclusive nodes. Each has 24 physics cores + 24 hyperthreads, so that corresponds to each hyperthread being ~45% of a physical core for these jobs. Sean will work on extracting useful numbers for calibration jobs.
      • There was talk of adding a "12 core" job queue based on poorly understood utilization numbers. No one knows the current number of machines that are configured to be in which queue. This should be on the SciComp website. Mark I. also pointed out that the usage numbers in the email he forwarded should be available to everyone. This could help us monitor our own usage.
    • Sean asked if anyone had looked at the mode 8 data that we've taken yet (planned as ~5-10% of the total production data). This data was to be used to check our reconstruction. No one had, though there is some confusion between which runs are mode 8 production data, and which are used for trigger studies (which often have detectors taken out of the readout).
    • Mark I. pointed out that on the offline FAQ, there is a SWIF section, which contains some useful information (including undocumented commands!). He encouraged people (especially Paul and Sean) to add to this.
    • Mark I. has tagged a new version of sim-recon
  • Calibration Train


Larrieu