Project

General

Profile

11/05/2015

Attendance

Matthew, Jeny, Vito, Joseph, Alex, Chris, Jianming, Jonathan, Satish, Ruth, Michael, Craig, Ryan, Ken, Neha, Gavin, Gareth, Bruno, Mat

News (Matthew)

  • At the Friday conveners meeting a timing shift bug was reported. This requires no action yet from production but we should be aware of this discussion.
  • We'll keep stand up meetings going through this week.

Simulation status (Gavin)

Gavin reports that he is currently about 50% done. However it has taken a long time and been really painful to get this far. The main causes of issue has been that the genie reweight module is taking >90% of the run time of the jobs and is pushing a large fraction
of these over the 24 hour limit (resulting in them being killed).

After some discussion we converged on the idea of not continuing in this manner. Instead Gavin will restart the remaining jobs without the reweighting in. And instead we'll run this at CAF time - the exact difference this will make is not entirely clear to me, but may result in the end user being unable to reweight rock muons. He'll make sure to flag the files without reweighting via their metadata so the two samples can be disentangled.

Chris agreed to look into the prerequisites of re-weighting at the CAF stage.

Reco status (Joseph, Satish)

  • ND non-stagger data is done.
  • ND stagger data is done except for one file, which failed due to a BPF error. We'll just drop this file for now.
  • FD data is running, slowly adding nodes. Josephs best guess is around 10 days to complete this set. The FTS is the bottleneck here, we should contact Robert Illingworth to ascertain if there is anything we can do about that.
  • All MC available is complete, with the exception of Tau's that Satish will now start slowly.

LEM status (Chris)

  • Chris reports that all the available files have been LEMed although some of the ND data files haven't yet shown up in SAM. He will look into this.

Hough vertex

  • See Chris's email.
  • This is being run in production but isn't exposed in the CAF. As analyses don't like mixing CAF versions (blame ROOT) Chris proposes that we hold off on committing the changes needed to expose this until the full set of files are complete.

Mix/CAF status (Bruno, Gavin)

  • ND data: Non-staggered are done (ones that are LEMed), staggered still have a bug.
  • FD data: ready to go, and will start ASAP.
  • MC: Done.

What if we wanted to top up with new data (Matthew)

If we were to top up with all the new data recorded to date this would mean the following be included.

ND

nd = sam.listFilesSummary(dimensions="data_tier artdaq and file_type importedDetector and online.detector neardet and online.partition 1 and online.stream 0 and nova.standard true and online.totalevents > 0 and online.runnumber > 10824")

{'file_count': 1435,
 'total_event_count': 3449818.0,
 'total_file_size': 67149133724.0}

That's 62 GB and 3.5 M events. For reference we currently have: 214 GB, 7.8 M events. This is a small dataset so adding to it won't be a problem. However expanding the MC by a commensurate amount would take a long time (1 kCPU day with the aforementioned dropping of the reweight).

FD

fd = sam.listFilesSummary(dimensions="data_tier artdaq and file_type importedDetector and online.detector fardet and online.partition 1 and online.stream 0 and nova.standard true and online.totalevents > 0 and online.runnumber > 19096")

{'file_count': 21604,
 'total_event_count': 3101207.0,
 'total_file_size': 1277780262061.0}

That's 1.16 TB and 3.1 M events. For reference we currently have: 6.51 TB 14.3 M events. This is a large dataset. At the current rate this will take around 6 days to reco/pid alone.

Notes

Other things that need to be done for this are:

  • raw2root needs to be run on all of these files.
  • Have to re-iterate through DQ and have to worry about bad channels…

ND branch discussion

See email from Mat. This discussion will continue in a reco forum.