Reprocessing: Meeting of 28-Feb-2005: 9:30-10:30 ESNet video conference

Agenda

1. News
2. Status of central production certification.
3. Status of Merge Certification of Sites
4. JIM Deployment and Remote Setup
5. AOB.

Minutes

Participants

Topics

1. News
   
   • p17.03.01 tar-ball available since last Tuesday (d0repro only). Copyd0om tar-ball to be published separately.
   • The d0repro tool is updated to reflect these changes, to be released when tested with the copyd0om tar-ball.
   • In connection with an improved file fetching logic in merging this should solve scratch space issues. (To be released)
   • Farm failures last Monday were attributed to changes in the calorimeter database. Caches might become inconsistent. Details need to be investigated and might need changes in the client code. Thus a new release of d0reco p17.03 might be needed. Will be discussed in tomorrows calibration meeting.
   • In general p17.03.01 is running much smoother than p17.02.00.

2. Status of central production certification.
   RecoCert p17.03.01 is crashing on clued0 in the first event. It sometimes works on CAB [Addition after the meeting: small filelist seem possible, DW]. Christian Schmitt did an effort of recompiling and running RecoCert. Though this version doesn't crash it is of no use as the resulting root tuples produce empty plots (or such with entries only at a value of 0).

   We currently have no way of running RecoCert in a defined way on the full certification dataset. Merging root-tuples from many separate runs is known to cause problems in p17.02.00.

3. Status of Merge Certification of Sites
   
   • DØFarm: Plots
   • WestGrid: Dataset of unmerged TMBs supplied and processed by RecoCert. (Awaiting end of shutdown to merge)
   • Lyon: Plots
   • SAR: Plots
   • GridKa: Dataset of unmerged TMBs supplied. To be processed with RecoCert.
   • Wisconsin: Dataset of unmerged TMBs supplied. To be processed with RecoCert.
   Differences on page 28 don't make sense, because plots of merged files a look more correct than those of unmerged files. We can't decide with certainty whether the differences come from RecoCert or are caused by a bug in copyd0om or its setup.
   

4. JIM Deployment and Remote Setup
   • Prague: Working on merge certification, still having weird errors. CRC checks on imported files were OK. Efficiency is still very low. Datasets for merge certification expected within the next days.
   • UK: IC: Debugging SAM station. Fix expected in the next days. Then JIM tests can continue.
     Manchester: JIM seems working, final tests ongoing.
   • CMS Farm: (No answer from Joel Snow -> Ask Amber)

5. AOB:
   --

Action Items:

Next Meeting