Core Minutes 1 October 2013

Core Minutes 10/1/2013

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

ScienceTools: (Jim) One backwards-compatibility issue affecting old bexpmap files has been fixed.

See Science Tools Development Notes for more information.
FSSC: (Joe, Liz) Fastcopy machines will stay up and data will flow to SLAC because it is essential to health and safety. Those involved may check NASA email.
Pass7: (Tom G.) Reprocessing Backfill complete through 9/17; will do another soon.

Comparison (Tracy) has been comparing GR 17-35-24-gr17 and the L1 candidate 17-35-24-lp57. Two possibly significant infrastructure changes (in Gaudi and in OS/compiler version) but (Leon) no signficant code changes found so far. (Tracy) There are significant differences in individual events, and not just a few events. Differences seen in CAL results are consistent with round-off, but the same can't be said for TKR. Thanks to heroic efforts by Heather and Joanne he is now able to look and make comparisons with the event display. There are different reported hit positions. This seems to point to some alignment (including alignment calibration) or geometry issue.
Pass8 reprocessing: (Tom G.) There have been more than the usual number of batch system problems. At 18:37 yesterday 300 jobs crashed all at once. Some hours earlier all jobs submitted to a particular machine failed. But, even so, we're making good progress: 2/3 complete. ETA is late November. (Tracy) is happy with the P8 production. People are using it. (Tom G.) astroserver is almost ready to go.

Preparing for L1current (Tom G.) We'll need to pick up new ACD calibrations when we get to that point in the data. Either we could use the override feature or else make a new GR with appropriate job options. [see discussion of new tags below in Core talk section]
Hardware: (Richard) Our new fileserver has arrived and is installed, prior to disk arrays. so should be available soon.
Core talk:
- GR tags (Heather) In addition to the jo changes needed for L1current there are the recent ACD (support structure) changes to the geometry. Should these go in at the same time? (Leon) Philippe has found optimal values for parameters affecting the structure; it's still being looked at. It has much more of an impact on simulation than on reconstruction. We can start with a GR incorporating only the L1current changes. (Heather) will make that tag, and another later with the geometry upgrade.
- Rhel6 validation (Heather) would like to get going with Pass8 validation (can't run L1 on rhel6 because it needs an old version of G4 which is not supported there). Will continue to use rhel5 for building. (Warren) We can't run on rhel6, even with rhel5 build, because currently ISOC code won't run there. He will ping Jim P. about a new build.
- Mac Pro (Tom S.) gcc version we wanted is installed. and there are personal accounts on the machine. He tried to build QT (RM code depends on it) and ran into some problems but they appear to be surmountable. The first one was that the versions of QT we use (newest being 4.7 on Windows) won't build on Mountain Lion because they're too old. 4.8 is supposed to support Mountain Lion, but it also failed to build because of a missing library. However the part that failed to build is a component we don't need. The build can be configured to omit it. He's starting with a 32-bit build of evrything; later we'll probably want 64-bit.
  
  We still need a glastrm account there. It should not be passwordless, but will need to use ssh keys to log in to the Linux passwordless account. Someone needs to set that up for us or reinstate password briefly so we can do it.
- Jenkins on Linux (Tom S.) We need a dedicated box or boxes to run on; then he expects it will be straightforward. Once it's all working he could strip all dependence on lsf out of RM.
- Release diffs (Heather) Would like to convene a meeting — possibly as part of this meeting next week — to settle on requirements. It should include interested parties like Leon and Carmelo.
- setupEnvironment (Joanne) made a new tag of facilities which eliminates the need to call commonUtilities::setupEnvironment explicitly in most cases. She does not recommend eliminating existing calls. They're harmless, and in rare cases the call is still necessary (tedious details available upon request). (Heather) This issue came up because of an attempt to build a new version of fitsGen. There seems to be no place where such procedures are documented. They should be.

minutes index

J. Bogart, Last Modified: