Dominick, Susan, Nick, Michael, Nate, Adam, Gavin, Jon, Eric, Ryan, Kanika, Chris
Dominick - No reconstruction processing since Monday
Crashing on certain connections
Web server is supposed to cache same query, act as a guard dog to db. Identical queries are held
If query is long (atten constants is 2M rows so it takes an age). Second query deems connection as stale so reconnects again so every new connection duplicates and so DB access is not working. Igor put in a new backend.
Started jobs 20 minutes ago.
Prioritise numi trigger files until Monday for neutrino search
nova_medium_prio - beaten into the ground can use nova_high_prio at own discretion.
Gavin - Reconstruction of MC is complete, at least for FD.
Satish has joined the production group! Thanks. He will take over PCHit generation.
Dominick kindly agreed to help get Satish up to speed and will process any pchit files in the interim.
Dominick: CAF, modified cafmetadata extractor to output json. Done, extractor lives in bin. Ready to go.
Action Item: Gavin will setup the FTS on vm03.
Maybe two versions of cafmaker can be run so that the blind/unblind files can be made in the same job. This works according to Kanika (she's successfully done it previously).
The metadata piece nova.special: "none" HAS to be set for all "DATA" dataset definitions henceforth.
Question for Andrew: Can we find it as a negative parameter?
Michael G. could set up the web GUI so that a the nova.special parameter is set to false.
No equivalent to reconstructed.base_release for caf.base_release (Adam can make that!) DONE.
ND Rock Generation¶
Rock generation: Initial round of rock gen problem, didn't set right output file. Files didn't come back.Fixed.
High failure rate from CMap problems. Nate, ran an independent sample - had success on 58/64. All failures due to CMap. Issue 6018 logged.
Action Item: FTS not setup for ND rock. Gavin to configure it for mcdaq dropbox.
Eric: Gen time is 330mins - 5 hours. 14MB, rock events, filter out empty spills. About 800 interactions in final rate, 5% efficiency for running. 1350 files , ~600. ND secondaries
ND overlay step testing from Nate: nova_sim pinged for help on the CMap error
Idea to use dedicated sites in future. 40GB total. Get data storage locations.
Action Item: Produce PCHit files from ND cosmics.
Overlay metadata: overlays, generate detector-only (has g4 info) overlay then do photon transport and simulation. Results in a second file (g4.root and daq.overlay.root) and both to get declared to SAM. Define how want to do parentage. Marc thinks we should set parentage for both files, in different tiers, to be the fcl.
Technical limitations that Adam described - I have no idea how to summarise this.
Essentially we decided dropping the g4 files (not saving it at all) is optimal and giving the fcl file as the parent to the final overlay file.
Action Item: Adam to put in a switch for input file parentage choice
Dominick: Moving towards the tutorial, can we move to art_sam_wrap.sh from ifdh? Right now we use our own.
Likely for Monday we have to use our own.
We should all use the development version.
We stripped /usr:/usr/bin from the front of $PATH and also we also removed process_name to use the basename.
Dominick submitted jobs, retrieval times are long - raw files are on tape. Things falling off of dcache . Dominick to submit a service desk ticket.
Quick tutorial session. We will likely meet up tomorrow as and when folks are free.