LHCb Condition Database Meeting

Date and Location

05. July 2006
11:20 - 12:20
CERN (2-R-030)

Attendees

Marco Cl. (minutes), Nicolas, Philippe

Subjects

Outcome of last COOL meeting

Marco Cl. showed few slides during the last COOL meeting.

The main outcome of the discussion was that we must have the possibility of using concurrently more than one COOL database instance. The current implementation of CondDBAccessSvc allows to connect only to one COOL database and the layering functionality provided by CondDBCvnSvc does not fit for the production usecase (too slow).

The reason of having more than 1 db instance, is the cross replication. Currently, different COOL folders do not share tables, but it may change in the future. The only way to ensure the usage of different tables is to use different COOL "database name" (or different schema). The cross replication via Oracle streams, may need to use two different schamas (schema owner accounts): one fro the replication from CERN to PIT the other to replicate from PIT to CERN.

Marco Cl. will try to implement a sort of delegation mechanism in CondDBAccessSvc that selects the COOL instance to use based on the path of the COOL folder. E.g.:

  • search for the folders and foldersets under "/onLine" in the database ONLINE
  • search all the others in the "default" database (backward compatibility)

Nicolas suggested that we can use the multi-db feature to have a separate DB for MC specific data (simulated on-line conditions, MC specific data). We agree that, once the new feature is available, we can do that for visualization data too.

Database status

Data

Marco Cl. started to port XmlDDDB v30r8 to an SQLite file. The big surprise is that XmlDDDB v27r1 (with 4 versions + testbeam and scripts) contains 1294 files, while v30r8 (1 version + testbeam and scripts) contains 3222 files.

This is a problem because the it means a lot of tables and a lot of overhead in the DB access. The number of tables can be reduced by using COOL channel ids. The most efficient solution is to merge many XML files into one (one round trip to the DB instead of many). In some cases it is not possible (like TT modules structure), but in others it should be done.

Philippe suggested to present the issue at the T-Rec meeting because the changes should be agreed with detector experts.

A.O.B.

  • Marco Cl. will try to start an EDMS document to collect all the main CondDB informations.

-- MarcoClemencic - 11 Jul 2006

Edit | Attach | Watch | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r1 - 2006-07-11 - MarcoClemencic
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback