Pass7 reprocessing:
(Tom) does the play-by-play:
- He performed a test last weekend consisting of approximately
10% of the data (400 runs, roughly the month of August 2008) using
GlastRelease v17r24. One run failed because the number of events in the
digi file did not match the number of events in the merit file...as
expected.
- Validation is underway and one GRB reprocessed looks unusual, so
another eight GRBs spread out over the entire 9+ months are being reprocessed.
- The pipeline task to perform merit reprocessing will be ready for
production once the final GlastRelease and jobOptions, etc. are approved.
-
A second task generates photon FT1 and LS1 (and expanded FT1 file with
additional merit variables), and will eventually generate electron FT1
files. This task is awaiting a Pass7 version of evtClassDefs and
ScienceTools.
-
During the weekend test, the pipeline experienced a burst of unexpected
problems related to the disk being used for batch job working directories
(/u18). Current suspicion is that this is related to a backup scheme on
that server and a change/upgrade will be (has been?) performed to mitigate
the problem.
-
Current estimate for merit reprocessing is not-less-than 1.6 days elapsed
time. That ideal situation will almost certainly not occur; 3-5 days is
more likely.
Pass7 reprocessing validation:
(Luca, Anders,...) Luca gave a presentation on what's been learned
so far. A comparison of sky maps for Pass6 and Pass7 reveals marked
differences,
still to be explained. (Richard) suggests looking for a problem
in transforming local coordinates to celestial.
See also a
first look at
Pass7 irfs.
- GlastRelease : (Heather)
There have been a few tweaks to ntupleWriterSvc over the past few days,
the most notable being a modification to RootTupleSvc::getItem so that the
output tuple is checked if the input tuple has a disabled branch which is
requested by the client. Unfortunately the version included in GR v17r25
does not allow for the standard case where there is no input ntuples, so a
patch has been applied. GlastRelease v17r26 is now being built with the
most recent patch to ntupleWriterSvc.
-
ScienceTools: Jim presented
the report.
(Richard) When will there be a ScienceTools release appropriate for
Pass7 FT1? (Jim) Within a week or so. Some other things need to go in first.
Core
talk: No Core Talk today. SCons meeting will take place
as usual Wednesday at 9.
J. Bogart, Last Modified:
01-Jun-2010 15:47:24 -0700