Select GR (20-08-12) to start off, but will change
GR now built for rhel6-64, but not validated - revert to rhel5-32
(arbitrarily) select ST (09-31-02) to start off, but will likely change
ST now built for rhel5-64 and rhel6-64 - not certain which should/must be used...
Discuss w/Yemi about generalized scratch space in batch farm, and short of that, the best way to determine whether to use /scratch or /lustre/ki/pfs/fermi_scratch
Get working jobOpts file
combine initial templates from Tracy/Johan (Pass8) and P202-ROOT
correction for need to produce GCR files (remove Cal3.Members={};)
correction for need to produce MERIT (not PRIME) ntuple
corrections for new ALG naming (e.g., PtValsAlg1 and AcdReconAlgV2)
gitify P300-ROOT and commonTools working directories for better backup, tagging, etc.
Create new version of commonTools, 00-02-00 for Pass8
Update commonTools/setupGR.sh
replace rhel4/rhel5 with rhel5/rhel6
Update commonTools/setupSkimmer.sh
Select rhel5 only (must be compatible with TSkim, ROOT, and GR rootData.so files)
TSkim v08-02-05 (rhel5-32 and rhel6-64 available...no rhel5-64)
ROOT version v5.34.03.gr01 (use rhel5-32 build)
fix problem with $ROOTSYS directory structure (for releases beyond 5.26)
fix problem with specification of file paths to search for rootData.so files
Update .xrayrc to use heasoft 6.13
Update commonTools/setupFtools.sh
remove force to 32-bit tools (no longer needed)
update repTools.py
clean up code
Update genRunList.sh and generate new list of DIGI files to reprocess (only first handful)
Improve/debug ability of Pipeline task to limit the number of events reprocessed per run and per clump
Add feature to config.py to call python's resource.setrlimit() to limit core dumps to 512 kB
New repro configuration: like P202 but without CAL and with RELATION files being produced (and including TSkim file prefix and merge dict)
Remove various ROOT-related hacks in P202 which worked around now-obsolete problems in v5.24
Modify GPLTools to use proper scratch space, i.e. /lustre via infiniband on the bullets, if available
Still to do.
Tune clump size for xlong queue
trade off more shorter jobs vs (increased LSF and DB transactions, and horrible response of Pipeline web interface)
Tune trickleStream for max number of parallel jobs, watching NFS and xroot behavior
Clone/update scratch cleanup mechanism
Clone/update removal of old (non-Pass8) RECON/CAL files
Validate sIntent query criteria, then generate full DIGI list