Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Procrastinate

procrastinate.readthedocs.io

  • Free plan
  • Windows
  • macOS
  • Linux
  • API
  • Self-hosted

What Procrastinate is

Procrastinate is a free, open-source task processing library for Python applications that defer work to background workers. It relies on PostgreSQL 13 or later to store jobs, manage locks and dispatch tasks. Jobs can run at a future time, after a delay or periodically, and tasks can use retry limits, exception filters, and constant, linear or exponential delays. Jobs sharing a lock run one at a time, ordered by priority and then creation time. The library supports synchronous and asynchronous code, with asynchronous use recommended when possible. Its Django integration can use Django database settings, load tasks from apps, and provide models, admin pages and migrations for Procrastinate tables. Command-line tools offer health checks and an interactive shell to inspect and administer jobs, tasks, queues and locks. The shell is described as experimental, and an administration web portal is not yet available. Windows is known to work, but the documentation advises Unix-like systems in production because workers may stop abruptly on Windows and automated Windows tests are absent.

Who it suits

Procrastinate suits Python applications that need background job processing, including Django and ASGI projects. It is for teams able to self-host and run PostgreSQL 13 or later.

Good to know, and what to watch

  • Supports scheduled and recurring jobs.
  • Offers configurable retry limits and delay strategies.
  • Django integration includes models, admin pages and migrations.
  • Command-line tools include health checks.
  • Requires PostgreSQL 13 or later.
  • Administration shell is experimental; no web portal is available.
  • Documentation advises against Windows for production.

Procrastinate

Free

Open source

The verdict

Procrastinate provides scheduling, retries and job administration for Python projects using PostgreSQL. Its experimental administration shell and Windows production caveat are important considerations.

What the maker publishes

Database requirement
Procrastinate relies on PostgreSQL 13 or later to store and manage jobs. source
Django integration
The Django integration can use Django database settings, load tasks from apps, provide models and admin pages for Procrastinate tables, and supply Django migrations. source
Error reporting
When a job raises an error, Procrastinate logs the error with `exc_info`; the documentation names Sentry logging integration and Google Cloud as tools that can collect these errors. source
Execution modes
It supports synchronous and asynchronous code, and the documentation recommends asynchronous use when possible. source
Installation
The quickstart installs Procrastinate in a Python virtual environment with `pip install procrastinate`. source
Intended users
The documentation presents Procrastinate for Python applications that need to defer work to workers, including applications using Django or ASGI frameworks. source
Locks and priorities
Jobs sharing a lock are processed one at a time, ordered by descending priority and then ascending creation time. source
Monitoring
The command-line interface provides health checks, and its interactive shell can inspect and administer jobs, tasks, queues, and locks. source
Monitoring limitation
The documentation says the administration shell is experimental and that an administration web portal is not yet available. source
Purpose
Procrastinate is an open-source distributed task processing library for Python that uses PostgreSQL to store task definitions, manage locks, and dispatch tasks. source
Retries
Tasks can use retry strategies with retry limits, constant, linear, or exponential delays, and exception filters. source
Scheduling
Jobs can be scheduled to run once in the future or after a specified delay, and tasks can also be launched periodically. source
Windows support
Procrastinate is manually tested and known to work on Windows, but its documentation advises using a Unix-like system in production because workers may stop abruptly and Windows has no automated tests. source

Checked against the maker's pages on 9 Oct 2026.