Project

General

Profile

11/17/14

Matthew, Bruno, Craig, Nate, Dominick, Nick, Susan, Chris, Ryan, Kanika, Jim, Jeny, Adam, Jon.

News (Matthew)

  • Matthew is back from his holiday.
  • Thanks to Dominick for running the meeting last week.

First analysis status and priority recap (all)

  • FD data - done through reco, no reason first pass PID and CAF can't start now (this has just been lower priority than some other samples), Dominick will do this.
  • FD beam MC real conditions - all done.
  • FD beam MC 14db - FHC done, RHC need to be remade because of a bug which meant that the old v04 flux files were used (the ones with the missing triangle of neutrinos) - these are still being generated.
  • FD cosmic MC - 2.8M done - What about the high statistics MR brem request? It was noted that the deadline for this is rapidly approaching. Nate mentioned that if we want to top up this sample then we need to code in a sub run offset (which doesn’t need a back port) and that he’ll need to switch to job sub client to get his priority working. - Matthew will communicate with Nitin, Duyang and the Nu_e group about this.
  • ND data - A problem was discovered last week where the ND data files were being reconstructed with an old geometry. Jeny re-raw2rooted these, and the resultant files have been processed up to CAF in Dominick's latest rapid turn around sample. While doing this a bug was spotted in calibration which leads to infinite energies when there is a hit in the muon catcher x-plane and the W is reconstructed outside the detector. A fix to this in LEM was hacked in rapid turn around. Alex R. is working on a fix which should go into calibrator today. When this is in we'll have to re-tag then re-run reco.
  • ND beam MC - Simulation is still ongoing for these, so far 34/40M events have been made. The available samples have been reco/pid/caf-ed.
  • ND cosmic MC - These have been simulated and turned into calibration files. Reconstructing these is currently low priority.

ND with calibration (Dominick et al)

See above.

ND geometry issues (Ryan et al.)

At the moment the raw2root stage invokes the geo service and this brings in the geo file and bakes it into the file. This methodology was chosen as it is inline with when geometry is included in the MC files. However, as discussed above, this can cause issues where an outdated geometry is used in raw2root as was done for ND. According to Jim this is not currently an issue for the FD - as the geometry hasn't changed recently, only the ND.

The ensuing discussion concluded that doing this at raw2root is not ideal and in future it would be good to drop this from the raw2root stage and instead have a new module which runs at reco, before cal hit which invokes the correct geometry. This will be implemented in the future,

RHC MC re-processing (Nate)

This will happen as soon as job sub client is working for Nate. When this is true he'll remake the files and retire all the children of the old set including the weird broken pidpart files Dominick reported finding.

Spill level DQ in production (all)

This module is still under going its final checks before being ready for production. Eventually we will want this in and it will run at the CAF maker stage. Its remaining changes should be ready in a day or so. These changes could go in a tag with with calibration muon catcher fix on roughly timescale of a day. It was noted that the CAF filling logic needs to be implemented. Jim will contact Louise about this.

Jobsub client bug (Jeny)

Jeny thinks that this was a temporary thing to do with timeouts as it has gone away now. Jeny also has a concerns about raw2root as some jobs are taking a long time and staying in a running status for a day. She is tracking down the bug to find out what is happening. Dominick says he has a ticket open about this and it’s got no traffic. Jeny will push on.

Reco keepup plans (Jeny)

Are we ready to turn this on? Matthew will contact DQ to see if they're ready to go.

Increase in sub run size plans (all)

The FD subrun size was increased by a factor of two on 13/11/14 (that's run 18162). Note: this isn't listed in the stable partitions page. Matthew will collect the information on which runs this is and then put together a short full scale production to confirm that everything works. On a longer timescale pieces will be in place so this can be doubled again.

AOB

  • Gavin - announced the 14DB sample through to CAF, geojittered FHC are announced.
  • VMs 01-03 are still screwed up after the slf6 upgrade. Gavin and Craig are worried that they’re going to upgrade everything and break everything. Craig will follow up with Andrew.