David, Sasha, Jack, Paul, Marco, Rob, Richard Hansen (welcome), Ed
Tier2 and general discussion
- We discussed differences in roles between US and LCG Tier2s (Sasha's points).
- Discussed DSS project in general. Agreed that focus on which data services should be available at Tier2 is still right direction. We need to flesh out the specifics.
- Rome datasets are in TAG databases (Oracle and Mysql) -- identical content
- BNL db is corrupt, don't use it.
- Collection schema description? POOL tools work with both (eg.
list-attributes) that come with ATLAS releases. pool.cern.ch or ATLAS database twiki.
- Are we routinely producing tags as part of AOD.
- pathena can take TAG dataset as input and process events, at least at BNL.
- it can take a tag file as input and turn it into a set of jobs
- How modular is existing code?
- PAT group is promoting pathena
- concerns about dependence on central services and complications
- Use CSC11 datasets and their associated TAG datafiles.
- Want to make sure we choose the right datasets to work with (not Rome datasets)
- Need more merged AOD files (and TAG files). An example file was 885 MB.
- Note: EDM has evolved in CSC12.
- Schema - there is the underlying structure (pool defined), and then the attributes (its in the ATLAS release). Jack will provide some documentation.
- Where does one go to find a list of tags?
- Building and updating a local tag database from tag files (Marco, Jack)
- Take a relational database, do query, then create output for something downstream, eg. pathena (factorize this).
- Find out which AOD files correspond to a tag database, including any query.
- Find out if referenced files in fact exist locally.
- Return information on the LFC, and the DDM for datasets.
- Capturing output of skims in a local database - Ed's suggestion.
- Factorize project into:
- Front-end: engine to query relational database and create tag files for analysis
- Back-end: modular framework to use any available job submission tool, such as pathena, or a local job submission system
- Investigate pathena capabilities and report
Calibration and Alignment data challenge - Sasha
- November, given delays
- What exactly should be hosted there?
- Not sure what the calibration model will be. Note - all subsystems will use COOL.
- Not sure what capabilities and services will be necessary at Tier2?
- Offline database meeting focused on event store issues every two weeks. Next meeting is Wednesday, 8:15am Chicago time.
- Tuesday afternoons, 2:30pm, weekly at least until the SW week.
- 01 Aug 2006