Core Minutes 17 April 2012

Core Minutes 4/17/2012

Present: Joanne Bogart, Anders Borgland, Jim Chiang, Richard Dubois, Warren Focke, Tom Glanzman, Heather Kelly, Michael Kuss, David Landriu, Leon Rochester, Tom Stephens, Tracy Usher

Power outage: (Richard) Power was off around 22:30 on Thursday, back 10:00 Saturday. The systems we needed were back up by mid evening. glastlnx12 lost its disk, which was replaced; Tony had to rebuild the info on it. And glastlnx06 (Fastcopy machine) had TCPIP problems which Steve Tether resolved by 03:45 Sunday. Then L1 was up.
ScienceTools: (Jim) Added an option in binned analysis for phase-selected exposure correction. See Science Tools Development Notes for details.
Reprocessing: (Tom G.) was out of town. Warren was the unlucky back-up. First he had to contend with an nfs server problem (sulky46), then the power outage. At this point reprocessing is about 1/3 complete.

(Warren) We had a lot of xroot problems yesterday, so I stopped submitting jobs

(Heather) The new GR reprocessing tag: 17-35-24-rp04 did not in fact include the updated ROOT v5.26.00a-gl6. This new ROOT had one small change to the xrootd client which decreased the time out associated with things like xroot server reboots. Tom G has requested a new GR tag where we actuallly pick up the correct ROOT. This will be GR 17-35-24-rp05.
Tapes: (Richard) SLAC is moving to 5 TB tapes. The drives have been ordered. Fermi is sitting on some 1500 tapes and several projects are asking for short term loans so they don't need to buy more. 40 to HPS, 100 to backups. KIPAC will want 400. LCLS may want some. I presume they'll all pay us back. And we'll need to look at the timing for our needs.
astroserver (Richard) Brian V. had to update the Oracle client driver version from 10 g to 11g.

The astroserver web interface now lists P7_P202_BASE (corresponding to the Pass7 reprocessing now in progress) as an option.
Pass7 news: (Leon) continues to port P7 fixes into the P8 version of the code. In particular The code to do software truncation of the trunc64 data back to trunc14 was developed in the context of pass7, because that is where we'll be doing it if and when we make the configuration change. But for completeness and maintainability, it seemed like a good idea to make the same changes in the pass8 code, although we won't be activating that feature. I've modified the code locally and I'll be committing it soon.

(Tracy) The first of the engineering runs for truncation took place on April 12th; data are available. It may be that the flight software changes will take a while, both because JJ would like to be sure he understands all the issues before implementing anything, and because he may have other priorities. Let's hope it doesn't take too long.

Geomagnetic latitude in the FT2 files: In 2010 we attempted to add a sign to the geomagnetic latitude so that one could tell which magnetic hemisphere the LAT was in. The N<->S transition is not defined completely consistently in the code, so there is some flipping back and forth as the orbit passes within a degree or two of the magnetic equator. The difference between the previous calculation and the current one shows up when comparing pass7v6 to the current reprocessing. (The magnitude of geomaglat isn't affected; the only difference is the newly-added sign.)

The plan is to release the reprocessed FT2 files as they were produced, including the above difference, and then figure out what we want to do about it later.
Pass8 news: (Tracy) The main focus currently is on background rejection studies. A CalRecon refactoring may be on the horizon.
Core talk:
- Systests (Heather) Systests have been done for a GR tag on the P8 branch. THere were some minor differences, to be examined by experts. Perhaps more of a concern — the all gamma overlay test gets stuck, but this is not new with this release. It has been the case since v19r4p0.
- MySQL migration: (Heather) mysql-node03, destination for the calib dbs and mood mirror (and current home of SCons RM databases) currently at MySQL version 5.0, might or might not be updated to 5.5. Once this gets settled we can do the move. Current target date for this and for the move of other databases to mysql-node01 is end of April. (Tom) When the new calib database is in place but before the alias move, he plans to ause submission, then restart with one or two jobs to verify as soon as the new alias definition is active.
- Mac support or not?: (Richard) Don't know. There have been no more comments from the Computer Center.
- rhel5 (Heather) There is a new GR build on the L1 branch. She's hopeful it will be able to handle real data.
  
  For rhel5 the only version of the obf external available is B3-0-0, whereas L1 processing currently uses B1-1-3. In theory this should not cause any problems since filters have not changed, but it's still a concern for some people. Heather will inquire to find out if there are earlier (pre B3-0-0) rhel5 binaries available. We suspect not.
- New SCons version (Heather) Transition is now complete. RM is using 2.1.0 everywhere and it is now also the default version for developers using SLAC central Linux machines. The workbook has been updated accordingly.
- RM news (Tom) Not much to report; everything is working. There was one tagging failure, probably due to a collision between tagCollector and RM making a LATEST tag. He was able to track down the bug which was allowing RM builds to proceed even when they were destined to fail because of incomplete tags. A fix is now in place.
- B3-0-0 (Joanne) had left a little mopping-up to do. The export collection Tony W. had created was missing some headers we use. One was a file which JJ was loathe to make public. Instead he made an alternate public file including the reduced set of definitions we're dependent upon. Last week I finally got around to testing with JJ's new file and somewhat modified arrangement of headers in $GLAST_EXT/obf, consistent with the proposed new export collection. It all appears to be ok.
- GR vc90 (Joanne) is working on addressing a couple of the more unfortunate features of SCons-generated project and solution files. The first — planning nearly complete; implementation in progress — is to get the project files to do installs of headers, xml files, etc. as is done by the SCons build. Next up: create an "all" solution file so that, e.g., it will be straightforward to build and run Gleam without extra manual shenanigans in Visual Studio.
  
  See also the Confluence page Heather created to capture user comments concerning SCons and GoGui: https://confluence.slac.stanford.edu/display/SAS/Comments+from+Fermi+SCons+Users

minutes index

J. Bogart, Last Modified: