Core Minutes 28 May 2013

Core Minutes 5/28/2013

Present: Joe Asercion, Joanne Bogart, Jim Chiang, Richard Dubois, Warren Focke, Heather Kelly, David Landriu, Tom Stephens, Tracy Usher

ScienceTools: (Jim) There is a new caldb tag, adding a new version of P7REP irfs.

Jean reported a problem in Likelihood. Jim has a fix, to be committed and tagged soon.

See also Science Tools Development Notes.
FSSC: (Joe) Ready and waiting for new irfs; no other news.
Hardware: (Richard) A new fileserver should be arriving around June 11 [update: started arriving today!]. We also nee to replace the Oracle server by the fall.
Pass8 (Tracy) 20-08-14 was released last week. Tom's test reprocessing runs all completed successfully. The output, now including the ACD as it should, was larger and CPU time was longer than expected. Tracker recon accounts for about 1/3 of the time. Mostly it's pretty fast, but there was one 15-minute event. That event had about 1200 tracker clusters. In this kind of circumstance (lots of clusters, not much space between many of them) significant time could be saved by merging clusters so the combinatorics don't overwhelm. We also need to be thinking about the other 2/3 which is not due to tracker.

Stephan Zimmer has made standard datasets so we can check things like whether resolution is back to where it should be.

Some job options values need to be adjusted (e.g. to get more timing statistics). We should also try out LZMA compression now that Data Catalog can handle this format. Tom will rerun the jobs once this is all in.
Core talk:
- GR Rhel6 migration (Heather) GR builds and runs on rhel6 but it has yet to be validated. That entails first running sys tests (possibly with some changes for Pass8 to stay consistent with new variable names and so forth), then examining the results. We need to validate both Pass7 and Pass8. lsf9 may not support rhel5-32bit (does support rhel5-64-bit — which is no help for GR — and rhel6-64bit) and in2p3 plans to go all-rhel6 by October. (Richard) Traditionally, we have not made these deadlines. rhel5 binaries run all right on rhel6. We could keep rhel5 build machines but run jobs on rhel6, assuming we can move Linux RM to Jenkins.
- Release Manager and Jenkins (Tom S.) expects no real difficulties in moving Linux RM to Jenkins. It should be no more difficult for Linux than for Mac.
  
  There is no news from Tony about the Windows/Jenkins problem. Tom will nudge him.
- Windows and lsf9 (Heather) We have contacted Renata concerning the transition to LSF9 and its impact on Windows. Apparently, LSF9 supports Windows Server 2003 (which is what we run on glast-win04) as well as Windows Server 2008, 2012. Likely we'll move to LSF9 first and then consider an appropriate time to upgrade the OS on glast-win04.

minutes index

J. Bogart, Last Modified: