Glossary

Procrastinate uses several words with a specific meaning in the documentation and reference materials, as well as in the user code and its own code.

Let’s go through a few words and their meaning.

Task

A task is a function executed at a later time by another process. It is linked to a default queue, and expects keyword arguments (see Define a task).

Job

A job is the launching of a specific task with specific values for the keyword arguments. In your code, you’re mainly interacting with jobs at 2 points: when you defer them, and when their associated task is executed using the job’s argument values. The analogy would be that a job is an instance of a task.

Using this wording, you could say, for example: “there are 2 tasks: washing the dishes and doing the laundry. If you did the laundry once and the dishes twice today, you did 3 jobs.”

Defer

The action of instantiating a task and registering it for later execution by a worker.

Queue

This is the metaphorical place where jobs await their execution by workers. Each worker processes tasks from one or several queues.

Worker

A process responsible for processing one or more queues: taking tasks one by one and executing them, and then wait for the queue to fill again.

Sub-worker

In case of asynchronous concurrency (see Execute multiple jobs at the same time), there are sub-workers, acting like the Worker, except there are multiple of them. They are orchestrated by the worker itself.

App Application : This is meant to be the main entry point of Procrastinate. The app knows all the tasks of your project, and thanks to the job manager, it knows how to launch jobs to execute your tasks (see App).

Job Manager

The job manager responsibility is to operate on jobs in the database. This includes both read and write operations.

Schema

The schema designates all the tables, relations, indexes, procedures, etc. in the database. Applying the schema means installing all those objects in the database. An evolution in the schema (modifying the table structure, or the procedures) is called a migration.

This term is not to be confused with that of PostgreSQL. In PostgreSQL a database contains one or more schemas, which in turn contains tables. Schemas in PostgreSQL are namespaces for objects of the database. See the PostgreSQL Schema documentation for more detail.

Lock

When configuring a job using Task.configure() you can attach a lock to the job. Jobs with the same lock are guaranteed to be executed in defer order, and in a sequential manner. No two jobs with the same lock can run simultaneously. See Ensure jobs run sequentially and in order for more information.

Queueing Lock

When configuring a job using Task.configure() you can attach a queueing lock to the job. No two jobs with the same queueing lock can be waiting in the queue.