Kanika, Adam, Paola, Jenny, Dominick, Chris, Nate, Ryan, Pavan, Matthew, Craig
Near Detector Data¶
ND FTS is up and running, RAW files being registered to SAM & put on dcache.
Kanika to tag S14-07-31 and Adam to start raw2root processing running on FTS in S14-07-18.
S14-07-18 release doesn’t work on the OSG due to missing packages. It works on the VMs and Fermigrid. Oasis is busted atm, so we can’t make a new release anyway.
Watchdog page for Near Detector is requested. Dominick can get something up and running in
about a week.
Michael Gheith has been working on a production-style dashboard. Matthew is keen to roll
watchdog page into this dashboard. Dominick should work on getting some feedback to Michael.
Nate is looking into time tags in the MC. Time is a uint64_t, first 32 bits are on the FEB and represent 67s of data, second 32 bits are on DCM and represent the remaining time. Not Unix time - nova time. Need special functions to convert. In MC, uint 32 stored as uint64, so time can only go up to 67s. w/o rewriting the empty event module Nate cannot see how to pass this along. Chris & Ryan say this works in EVD, DAQ2RAWdigit etc, so just do that.
Start there, Gavin will provide a second pair of eyes. But we’ll need our own empty event builder Nate says. Chris says put an issue in. change the 32 to a 64. Nate thinks our own empty event builder may be the way forward. If we send it to them we may not get a rapid turn around on it.
Kanika can set it up in advance. And will roll out the schema next week.
Ryan put stuff into make sim fhicl to draw pot run number and pot based weighting. Needs some accounting.
Do we define our final dataset as just a long list of file names rather than being meta-data driven.
No fundamental objection just the scripting will be more painful.
Can we add this as a field into metadata? Can do this after files are in SAM. Could just add a for first analysis field.
Can we do it via weighting rather than throwing away events? Other alternative to to top up. Can weight by editing sum PoT?
Want to generate a factor of 1,000 more MC than data.
Followup with Joao necessary. Has what he needs to make a start.
Nate working on this. Not looked into it. Will run a job today of 100 events and see if it works.
Pavan gave us an update.
S13-10-11 of 5.1TB has been transferred to dcache.
Calibration files, so far transferred 10TB. After completed, can then remove them from bluearc.
Neutrino 2014 are difficult to work since SAM makes life easier. At some point should move to SAM. Setting metadata field to "neutrino2014".
Move PID files to SAM. Pavan will eventually do this. Not urgent.
Chris can't find final PID file after mixing. Gavin will investigate. Did I not do the mixing step??
CAF files exist for reco and full processing steps.
Gavin to complete this by Saturday.
Dominick FPE exceptions mostly due to fuzzy k, fixed in S14-07-18
Grid periodic remove still an issue, 10% failure rate seen. Chunked calibration loading helps. Suggestions for getting these back didn’t work. Try lowering the u-limit
Gavin working on making this more fluid.