Data Management meeting held the 30/09/2003 Present: P.Charpentier, J.Closier, M.Frank, V.Garonne, G.Kutznetsov, F.Loverre, S.Ponce, A.Tsaragorotzev Minutes and action list. Agenda: ======= - Fixing time slots for future meetings (Markus) The meetings will be every Tuesday at 9:30 Markus will arrange a telephone conference for this time slot. === Action: Markus - Bookkeeping - Metadata and job tracing Review current status (Sebastien) Sebastien gave an overview of the current system. Now all data accesses for the user interface are implemented in JAVA. The code still needs some cleanup to be done by Francoise. Also the implementation to select datasets not only by reconstruction and simulation program, but also by e.g. stripping program, event tags etc. must still be done by Sebastien. Once the implementation is finished, the JAVA code will move to the ORACLE web server, where Joel agreed to arrange the move. === Action Francoise, Sebastien, Joel - New ideas on retrieving the information (Andrei) The group around Andrei will investigate the use of the Alien bookkeeping system, which is based on a hierarchical structure. The idea is to move towards a standard ARDA interface. === Action Andrei - Filecatalogs - Technology choice (EDG-RLS, Alien, POOL-MySQL-LHCb) (Andrei) The file catalogue will not only have to be used to resolve physical file names from file identifiers (FIDs), but also to manage job distribution etc. Since this overlaps very much with what Alien intends to do, Andrei will investigate the possibilities of a possible implementation based on Alien. Any implementation must be based on an ARDA interface, which will have to be defined and endorsed by LCG. The file catalogue interfaces of POOL and ARDA will have to be compliant. === Action: CPPM/Andrei - User analysis and simulation productions (Andrei) - Workload management service(s) - Reprocessing and individual user analysis - Simulation production Andrei presented his ideas of a new work load management system. The approach is very generic and will also allow to manage user analysis. In this context the existing production system is only one possible client of the work load manager. Ganga and others will share the work load manager. The development of this system will largely be done around the group of Andrei and a prototype will be available soon. === Action: CPPM/Andrei - New production tools under development (Gennady) Gennady presented his ideas for the development of implementing a modular data production editor based on the work load manager under development by Andrei. The editor is split into several components, which deal with the development of modular processing steps, the construction of work flows and finally the incarnation of a work flow into a data production with many jobs as it was described in the data management paper. These tools, which will ease the work of the data production manager will be integrated into a DIRAC desktop environment for a coherent look and feel. This work was seen to be very important, because it will allow to also describe data reprocessing and data reduction tasks, which in the current production system cannot be handled properly. The general impression was, that the plans of Gennady allow for a very flexible implementation, which would also allow to deal with Ganga user analysis jobs. How this can be achieved will have to be investigated. Philippe and Andrei were worried, that parts of this work would duplicate the efforts of the work load management system under development in Marseille. Hence Gennady and Andrei are invited to coordinate their efforts and work together more closely. === Action: Gennady, Andrei - Necessary changes to the existing (All) production system Cheers, Markus Frank