Core Minutes 19 February 2008

Core Minutes 2/19/2008

Present: Joanne Bogart, Anders Borgland, Philippe Bruel, David Chamont, Jim Chiang, Richard Dubois, Warren Focke, Berrie Giebels, Tom Glanzman, Navid Golpayegani, Heather Kelly, David Landriu, Leon Rochester, Eric Winter

Chuck status: (Richard) On Friday Chuck had emergency surgery to remove his gall bladder. He returned home yesterday. Richard has received email from him (dated 3 AM! Chuck said he was still on hospital time). He seems to be in good spirits; apparently the operation is what he needed.
Big run: As always, follow progress on the Big Run Checklist. (Richard)
- we are starting the L1 stress testing. The input was completed with 13 failed runs out of 345,600.
  - a similar timing problem with FSW was found (and gotten around), so Anders is remaking the L0 files now
  - SCS is putting pre-emptable queues on 900 cores so that L1 jobs will start essentially right away after submission
- we will not do any more generation for opssim2; we'll use the first 7 days with the last 2 having non-standard calibrations in recon
- we need to start the disk cleanup to free 100 TB
  - 9 TB can be had from one of the Xmas big run attempts
  - Julie is looking at prioritising deletes, with 3 categories — complete nuke; keep merit; keep all.
Science Tools report: (Jim) discussed some of the items in the Science Tools Update for Feb. 19.
GR news: (Heather) The current big run GlastRelease tag is v13r9p4. A new v13r9p5 will be created today once the HEAD build including Leon's branch tags of AnalysisNtuple and TkrRecon completes. The Big Run tags are all created along the GlastReleasev13r9 branch. The intention is to keep the updates along this branch minimal. Meanwhile, development continues along the HEAD of GR. As Richard notes, we should become accustomed to working in this way to support the "pipeline version" of the code while allowing development to continue. The Opssim2 runs will be using v13r11p1 due to the desire to utilize the latest GRBCalib tag v1r3p3. This tag also includes the new trigger code put in by Martin to address JIRA TRG-12.

We should begin preparing for a new GlastRelease tag that incorporates the many outstanding tags that currently exist in GlastRelease HEAD. This should be a v13r12 tag. Leon's updates along the branches of AnalysisNtuple and TkrRecon should be merged into the HEADs of those packages and tagged for inclusion as well. As usual details concerning all GlastRelease tags can be found in Confluence.

See the list of tags included on the HEAD of GlastRelease, as of today. There are number of things there, most noteworthy perhaps are the new CAL and EbfWriter tags from Zach. This may finally solve our rhel4 woes and also includes a more realistic version of the CAL processing in EbfWriter. It would be helpful to make a list of high priority items to have included in a new GR tag, perhaps we can create such a list by next week's meeting.
Ground software freeze: (Richard) News from Steve is that the Mission is setting up an expedited CCB process with 24-hour turn-around — and that the answer to our requests will not automatically be "no". Development (HEAD) is not covered by this process, only code which is to be run in the Pipeline. The freeze will go into effect on 16 March.

(Joanne) What do we want to do about ACD geometry with trapezoids? It should not be incorporated into any Big Run activity, but preferably should be in some release before the Freeze. Versions of AcdUtil, AcdRecon, etc., in existing releases is trapezoid-aware but has not been tested thoroughly. Fred will draw trapezoids; the old gui will not. (Richard) This should be considered and coordinated with other upgrades which will stem from the ongoing ACD code review.

(Leon) What is happening with support for the old gui? (Richard) Ask again next week. Currently we have no support for Fred; that may be remedied. With adequate support for Fred we shouldn't need the old gui.
Composite event lists: (David C.) The transformation of a ROOT CEL into a TEXTUAL CEL (which can be used as an input to the skimmer), has proved more difficult than expected, especially the transformation of the tree entry numbers into runid/eventid pairs. A solution has been finally found and the implementation is under way in the class ReverseIndex and the application celConvert (package rootUtil).

The problem with the TVirtualIndex provided by the CompositeEventList is probably understood, and appears to be in the testing code rather than the source code. Please note that the run and event ids which are stored in many GLAST trees must always be in increasing order, or we could end up with tricky bugs involving CELs and skimming.
Skimmer: (David C.) has not yet had a chance to investigate why the Skimmer crashes when using ROOT 5.18, but will do so shortly since the move to ROOT 5.18 is highly desirable (starting from 5.18, we expect the organization of ROOT components/packages will be stable again). He is also starting to work on another task: add to the Skimmer the ability to receive a ROOT CEL as input.
Services lost: (Heather) is wondering about the impact of the proposed list of reduced or eliminated support services at SLAC. (Richard) This list never was vetted with the user community. He expects support may be found for some items if users can make a strong case for them. VPN is one which is important to us. Leon might squeak about losing it [(Leon) squeak!]. Tony and Max are dependent on it to access Oracle from home. The difficulty for SCS in continuing to maintain it probably has to do with the potential for security problems.
SCons status: (Navid) He's been making small bug fixes here and there. Thanks to help and feedback from Heather SCons on Windows is in much better shape. ScienceTools is just about working now on Windows as well as Linux. Navid is hopeful about Mac as well, but we need to revive the Mac batch farm in order to try it out with the new RM. (Heather) was pleasantly surprised by the completeness documentation provided by SCons Central and by their responsiveness. What is the status of the web interface for the new RM? (Navid) hasn't heard anything from Karen; will ask.
Many thanks to Richard, Heather and David C. for substantial contributions to these minutes.

minutes index

J. Bogart, Last Modified: 01-Jun-2010 15:46:56 -0700