Execute multiple jobs at the same time

By default, each worker only takes one job at a time. One of the ways to increase your processing throughput is to launch multiple workers, each using their own process.

But Procrastinate supports asynchronous tasks, and the main idea behind asynchronous code is to have concurrency: while a job is waiting for external I/O (usually network), the CPU is free to run other jobs. That is why Procrastinate workers are able to launch multiple asynchronous sub-workers.

You can do this either from Python:


Or from the command line:

$ procrastinate worker --concurrency=30

The discussion section contains a few important guidelines regarding asynchronous concurrency (see Asynchronous operations & concurrency).