Project

General

Profile

08/07/2014

Attendance

Eric, Nate, Craig, Chris, Ryan, Dominick, Jan, Kanika, Jon, Gavin, Adam, Michael, Satish, Jeny, and Paula

The plan is for Matthew to take over the chairing of these meetings after August 11.

Timeline for first physics reminder

Sim tag first of September…
Alternate geometry could be ready next week.

Action Item: Craig/Ryan will follow up with Simulation group. Discussion point for Monday's meeting.

Plan for MC generation

Eric is ramping down quickly to write his thesis.
Others will have to carry the load including Nate, Jeny and Paola.

Action Item: Eric will meet with the new CD production group. Time set at 1:30pm Friday 8th August with Nate also.

SAM metadata in new art version (Related to ND MC effort)

ND MC has the PID, but no PID files made it into SAM
Errors due to the new ART version and the metadata it is expecting.
Adam — working to improve the module to handle the new formats. He wrote it and tested it last night – needs to test it in real conditions.

Action Item: Tests today. Nate will roll forward the FTS, and setup a test FTS.
Hopefully this will move quickly.

For future, Adam recommends making our own service. We don’t want these things to change so much going forward.

ND keep-up processing

Adam worked on scripts S14-07-18 (has near det stuff required).Had to update the setup nova grid functionality.
Some jobs are failing in the new version of root (memory usage). Keep-up is running on ND – about 1000 files already there.
Processing everything.
Our limit is in the wrong base version.

Action Item: Gavin to announce availability of ND raw2root files to nova_offline.

Progress toward managing FD keep up

Jeny and Paola will meet with Adam next week. Monday?
Will work on monitoring tools – met with Eric today.
Gavin has added them to the mailing list.

Action Item: Jeny and Paola to meet with Adam on Monday for further knowledge transfer.

GRID processing

Andrew had a discussion with FNAL CD Security – no longer need user kerberos certificate.
novapro will have its own submission ability thus allowing users to run as novapro and themselves in parallel.

Action Item: Needs to be tested that this is a smooth, transparent transition as promised.

Status of GENIE+Cry overlays

Nate produced a test file out of the box. Seems to work — caveat that Susan’s test had a problem with the way she looked for MCTruth.
BackTracker should work fine, but “generator” v/s “cry” module labels. Need a few more tests…

Action Item: Nate to run through the full suite of reconstruction/pid/caf

Filter for empty events, in Filter package, changed from using “isRealData” v/s Backtracker for a file.
(should come up with a simple filter module — we don’t want to run BackTracker in the simulation job)

Action Item: Gavin will resolve this so sim jobs don't use the generic filter package and can avoid instantiating BackTracker.

Time Stamps for MC Generation

Done. Works. In test release.
Not perfect: Good enough for our current purposes.
make_sim_cl will pass in the correlation between run numbers and time
Bad channels relies on RunHistory and RunHistory needs a run number

Action Item: Ryan will make the changes to make_sim_fcl (done)
Action Item: Nate will present at the DetSim Monday's meeting.

Reco for FD MC and PCHit status

Satish says Reco jobs are almost done — some failures. Detailed email was passed around.
PID processing mostly done too — not sure if the crashes will be repeated.
PCHit production going fine. Now ND files are available. Change FarDet to NearDet.
Why not do this in attenprof? Not quite ready? Backport, or new tag will be required.
Back processing may be required… Do a small test sample first.

Action Item: Gavin to relay relevant information to Satish to get pchit generation running for ND.

Evan has timing calibration ready for production. Needs a new data tier.

Action Item: Adam to make a "timecal" datatier (done).

Branching Progress

Kanika has got the scripts together for merging nightly. Start testing soon.

Action Item: Kanika to let us know when branching is ready for further guinea pigs.

Recent experience with NuMI data efforts

Dominick presented doc-db 11780
Discussion on failure tests. Two failure modes: Floating point exceptions and memory spikes.
CAFMaker floating point exceptions seem to be in much better shape due to many fixes.
NuMI triggers seem to fail at about 20%, attenuation points in diblock sized chunks reduces failure rate to 1/1000 (this could get us through the first analysis).
Still working to understand the "periodic remove" memory issues.

AOB

New art release: artists bogged down with C++ summer school.
Addresses some memory issues – will come out next week?