Code structure - PanDAWMS/panda-harvester GitHub Wiki

Harvester code structure

• Body: Harvester agent modules (subclasses of thread) using code in their specific subfolder. Master is the main module spawning rest of agents.
• Config: Module to read configuration file
• Core: Central components (e.g. class for DB access, classes to represent files/jobs/events, class to talk with panda server) needed in other modules
• Cred manager: Module to maintain credentials like grid proxy
• Messenger: Communication classes to interact between workers and harvester
• Monitor: Classes to monitor jobs, e.g. through SLURM, SAGA...
• Mover: Base classes or tools for Preparator and Stager
• Preparator: Classes to prepare (stage in) data for jobs
• Stager: Classes to stage out data
• Submitter: Classes to submit jobs to the batch system
• Sweeper: Classes to kill and cleanup workers
• Test: Test scripts
• Worker Maker: Makes workers