NotesJan16

Attending

Jerry, Jack, Marco, Robert

Agenda

Phone conference ESNet Call: 510-665-5437 When: January 16, 2007, 02:30 PM America/Chicago Meeting ID: 22333

Agenda:

• Make the point of the activities
• DB access requests (what access, what is implied)
• ANL analysis workshop
• Plan of action
• work on a prototype

Discussion

Jack: pathena at tier2 and tier3. t is working OK, only the examples. Install all the panda tools including pathena. Everything is n a work area and tars everything in the work area across. Difficut to separate data from the executables or different users. Dataset required should be already registered in DDM, sice it is using dq2_get nad registers the output in DDM. If should not reqire thaht the dtaset is local. Jobs were submitted from the prototype Tier2 (tier2-06) and from the tier3 at ANL (atlas16).

Pathena used to be part of the release. Now you get it checking out the panda tool packets, you use gmake and it goes back into the release. Use CMT from CERN to get the physics analysis tool (panda tools + user analysis packages). CERN wiki: DAonPanda, pathena on MWT2 - contains intructions Pathena works on ATLAS grid

Pathena, is strictly tied to the DDM and has builtin naming conventions. Strips TAG from the AOD and it assumes that that is the file name.

Interface of pathena? usable? suggestion for skim-selector?

TNT - the front-end may be OK but really it may have not the capabilities of running on OSG. It has assumptions that are even more stringent than pathen. At this point it is worthy to use pathena (even if TNT has a similar architecture to DSS). LCG-centric, to have it working for OSG we would have to do a joint project. The conclusion is that TNT is not a viable option.

There is a lot to be said to have something lightweight, lighter than pathen to run in a local site. Compatible with pathena if we go outside from the side. Something that could run lightly on a tier3.

Tom, Eric, Ambrish. Running on UC-Tier3 (PBS scheduler)

Build a package with a single configuration file, no production quality required for this first version.

Fabiola: Use Giulio's page to look into the Chicago DB. Now she requested a CERN DB to be able to do the TAGs tutorials. Tutorials from physics analysis group (Ketevi)

Register the TAGs in DDM so that they may be able to use them with pathena.

Action items

• Ask Tom to useful datasets
• Jack and Jerry produce the TAGs of one of them documenting the process
  • building TAGs
  • loading TAGs
• Marco produces the TAGs of another one reviewing the documentation
• Packaging the scripts and provide a single configuration file (Marco)

-- MarcoMambelli - 16 Jan 2007