Core Minutes 6/9/2015

• ScienceTools: (Jim) No news.

  See Science Tools Development Notes for current activity.

• FSSC: (Joe) There is ongoing discussion concerning the release date for ScienceTools, which will likely be June 24th. This will coincide with the planned public release of P8 data.

• Reprocessing: (Tom) Backfill started last week which handled 1 month of data. This leaves 3 months of data to backfill starting June 24th. It is all proceeding smoothly. (Richard) How far along? (Tom) P8 Recon is done for the most part, current through last Thursday. Next up are the merit files, and then FITS files. (Richard) So the lion's share of the reprocessing is done? (Tom) In terms of computing cycles yes, not in terms of human cycles.

• Hardware: (Tom) The switchover to the new gpfs user disk is complete. Yesterday, the user disk did show signs of stress due to one user running 700 batch jobs. The user disk did not melt down. We are learning how to observe the behavior and how to diagnose problems. There is still tuning left to be done. This system is more complex than a single NFS server. Discussions are ongoing about yesterday's event.

  (Richard) In other news, we've ordered 1/2 petabyte of tape. No word on the order's status.

• Pass 8: (Leon) Nothing to report. The C&A meeting did incude a presentation about cosmic ray analysis. The discrepancy between Hess and Fermi data above 1 TeV persists, but it is difficult to decide who is right. (Richard)Any news about AMS? results? (Jeremy) Not yet. (Warren) In preparation for P8, they will have to tune the settings for the slower reconstruction via 2 parameters in the config file. The Recon runs 2.5 times slower. To compensate, they will generate more crumbs but the same number of chunks. (Richard) So the throughput will be the same? (Warren) Pretty close. We have plenty of cores, and we're not using all of them most of the time.

• XRootD: (Tom) Wilko is updaing the xrootd cluster before leaving for a month of vacation. They will be getting rid of the 4 oldest xrootd servers but taking their disks which are superior to the Seagate disks in the newer servers. Anyone who monitors xRootd may notice. There is also a problem when reading very small files from one of the gpfs? servers - they are hoping to learn more about the issue too.

• Core talk:

  • Wired and retiring glastlnx07 (Heather) With Tony and Dima's help, Heather can run Wired on Windows. She still cannot run it on SLAC's Public linux machines - but that is unnecessary to continue testing Wired with GlastRelease. (Leon) What about Fred? Heather did try running Fred on SLAC Public a month or so ago, and it failed, presumably due to a Ruby upgrade. Assuming a user runs using an older version of Ruby (1.8.6? or 1.9.2?) it should work. Leon suggests waiting until someone complains before expending any energy on it.

  • Doxygen Generation (Heather) Last week, Tracy via Richard, asked about our automatically generated Doxygen generation. The CMT documentation was no longer available via the web links and the SCons documentation was never web linked. The issue with the CMT documentation was due to the announced changes for the SLAC Web servers, where directories would not be automatically indexed unless they contained a special file that indicated they were meant for public viewing. Once this file was put in place, the CMT Doxygen documentation reappeared: CMT Doc The SCons generation Doxygen documentation is also now available: SCons Doxygen Doc However, it has been determined that the DOxygen under SCons has not been generated since last May, 2014. This was confirmed by also attempting to produce the Doxgeyn doc by hand via the SCons --doxygen flag. A JIRA has been filed and hopefully Joanne and Heather can take a look at it: JIRA GRINF-69

previous

minutes index

next

J. Bogart, Last Modified: