The PanDA Production and Distributed Analysis System has been developed by ATLAS to meet requirements of data-driven workload management system for production and distributed analysis processing capable at LHC data processing scale. PanDA manages both user analysis and production jobs via the same interface with processing rate from 250 to 300 K jobs
on more than 170 sites every day.
PANDA INSTANCE AT JINR.
Goal of the project was to evaluate the possibility of COMPASS jobs to be executed by any COMPASS user via PanDA server installed at JINR on local cloud service.
So, the main goal was split into the following tasks:
• evaluate PanDA for MySQL instance installation;
• evaluate PanDA monitoring application installation;
• extend PanDA server&pilot by modules to introduce COMPASS’ logic;
• evaluate whether COMPASS’ software is portable and can be executed outside CERN;
• test that COMPASS payload submission via PanDA;
• prepare and publish a wrapper to define COMPASS job via PanDA client;
• make sure that an ordinary COMPASS user can submit jobs via PanDA.
All these tasks described above were finished successfully. Three PanDA queues were defined:
1. at JINR on local cloud service,
2. at CERN on lxplus and
3. at CERN on lxbatch.
Several jobs were sent to each queue. During the evaluation more than thousand real COMPASS jobs were defined and processed by PanDA. Successful processing of the jobs was confirmed by COMPASS software coordinators.
PanDA simplified data f low presented on Fig. 2.
