Core Minutes 7/9/2013

• ScienceTools: (Jim) No news this week.

• FSSC: (Tom S.) No news.

• Pass7 Reprocessing: (Tom G.) No activity yet.

• Pass8 Reprocessing: (Tom G.) After one week approximately 1000 runs have been reprocessed. At that rate the whole thing should take about 6 months — not bad! Three kinds of problems have been seen:

  1. Problem in ACD code affecting a small number of the runs.
  2. Problem with calibrations, affecting 7 runs out of the total. [This was caused by a small time gap in calibration coverage. Leon fixed it shortly after the meeting.]
  3. SLAC-wide sporadic email problem affecting otherwise perfectly-good runs. The solution (until the underlying email problem is fixed) is to roll back and redo all affected runs.

  Other than the email problem, infrastructure is holding up well. Tom will try cranking up the number of nodes from the current value of 2000 to 2500 or maybe 3000. (Richard) You can follow progress from the Pipeline pages.

• Pass8 (Tracy) There have been about 23 failures in the ACD code to date. The reason is that occasionally the first hit track coordinates are outside the geant world volume. When these coordinates are used in extrapolation an exception is thrown. Sometimes it is caught and handled properly; other times it causes a segfault. So the problem is understood; the tricky part is finding a solution which does not change results for any good events, so that we won't need to redo all runs which have been reprocessed so far.

  For the next few weeks up until the Collaboration Meeting, focus will be on analysis, beating down backgrounds, etc. Bill is looking at the cosmic ray tracker. His preference is likely to be running it all the time, even though it increases total cpu time by about 25%. Tracy leans towards running it just on a sample, as a cross-check. `

• Core talk: Waiting until next week when Ms. Core Talk returns from vacation.

previous

minutes index

next

J. Bogart, Last Modified: