Home

Welcome to the grid_simulation wiki!

Below you can see schematic representation of dependencies between functions.

1. Program starts to parse jobs.yml and then populates job queue.
2. Due to having free computing capacities, tier sends a job_request to a scheduler().
3. Scheduler selects (by using matcher()) appropriate jobs from the queue.
4. Then it sends a job batch back to tier. executor_launcher starts processes which will execute jobs from job_batch from scheduler.

5.Executor stage

• 5a. Executor starts to copy files from tape to disk if there are no available datasets on disks.
• 5b. Executor starts to download datasets from disks and reads them.
• 5c. Now tier is ready for executing task and it does it. Task produces output files at the end of simulation.
• 5c. Data_replicator launches uploader process which will replicate output files.

If anomaly occurs job will be rescheduled.