A lightweight monitoring and accounting system for LHCb dc 2004 production

The phase 1 of the LHCb Data Challenge 04[1] includes the simulation of 200 million simulated events using distributed computing resources on 63 sites and spanning over 4 months. This was achieved using the DIRAC [2] distributed computing Grid infrastructure. Job Monitoring and Accounting services have been developed to track the status of the production and to evaluate the results at the end of the Data Challenge. The end user connects with a web browser to Web-Server applications showing dynamic reports for a whole set of possible queries. These applications in turn interrogate the Job Monitoring Service and Accounting Database by means of dedicated XML-RPC interfaces, querying for the information requested by the user. The reports provide a uniform view of the usage of the computing resources available. All the system components are implemented as a set of cooperating python classes following the design choice of LHCb. The different services are distributed over a number of independent machines. This allows several thousand concurrent jobs monitored by the system.