Core Minutes 7/28/2015

• ScienceTools: (Jim) fixed a bug in the piecewise power law spectrum in Likelihood.

  The FSSC Ubuntu virtual machine mentioned last week can't be used outside Goddard; it's a dead end as far as we're concerned.

  See Science Tools Development Notes for current activity.

• FSSC: Joe couldn't make it to the (very short) meeting this morning, but reports

  We're hoping to get a patch release out mid-August, and we want to include the {ROOT/Python] update if it was ready in time.

• Reprocessing: (Tom) All is quiet this week.

• Hardware: (Tom) No news here either.

• Core talk:

  • Dropping like flies Platforms, that is. Jim has suggested we could do without rhel5, both 32-bit and 64-bit. Richard and M.E. have no objections to it. It would lighten the externals maintenance burden significantly. There are still issues with building CHS on rhel6, but we hope to drop rhel5 for all other container packages fairly soon. [But not until there is a happy resolution to topic below.]

  • rhel6 build failure (Joanne) Calibration was recently bitten by the leap second bug (see JIRA GRINF-74). It shows up only in rare circumstances which could be avoided by careful choice of calibration validity interval, but, nonetheless, it's a bug, so I made new tags of facilities and calibUtil to fix it. The rhel5 32-bit completed successfully (just a couple test program failures, similar to other recent builds), but rhel6 64-bit didn't get much past CVS checkout. The lsf batch job which should have done the SCons build apparently gave up very early. It turns out this is not a new problem. ST builds on rhel6 have been failing since July 21 with identical symptoms. The last good one was on July 17. The lsf log file from one of the failed jobs says "Cannot open your job file: /u/gl/glastrm/.lsbatch/1437498842.914711"; that is, the job, which is running on bldlnx04, apparently is not able to write to the directory it normally writes to. bldlnx08, which is used for rhel5 64-bit builds, is having no such problem. There was an outage on July 20 to upgrade lsf which may be relevant. See also JIRA RM-41. (Heather) will send a note to unix-admin. [Late update: Renata replied that the lsf daemons had not restarted on bldlnx04 after the outage. She has restarted them now. Heather will trigger another build. ]

  • WIRED on fermilnx-v14 (Heather) made some serious progress on this issue while at SLAC last week. It works! Meanwhile, she discovered an event id display problem: the event itself is ok, it's just the id which is not displayed properly. This has to do with different ideas of the type of this variable in newer versions of GR (long long) and HepRepServer. The latter needs to be updated. See details in JIRA GRINF-73.

previous

minutes index

next

J. Bogart, Last Modified: