dbos

What it is and what it does

DBOS is a Python library that adds durable workflow execution to applications by storing workflow state and progress in Postgres. Instead of managing a separate orchestration service or task queue, you annotate functions as workflows and steps, and DBOS automatically checkpoints their execution. If your application crashes or restarts, workflows resume from their last completed step without data loss or duplicate execution.

The library provides several Postgres-backed primitives: durable queues for background task execution with concurrency control, scheduled workflows using cron syntax, durable sleep for long pauses that survive restarts, exactly-once event processing for webhooks and Kafka consumers, and programmatic workflow management to query, pause, resume, or restart workflows. It targets applications that need reliable failure handling—payments systems, data pipelines, AI agents, and long-running background jobs—without the operational burden of external infrastructure.

Use it for:

• Build a payments service that reliably processes transactions even if servers crash mid-operation, with automatic recovery from the last completed step.
• Create a data pipeline that checkpoints progress in Postgres and resumes from interruption points rather than restarting from the beginning.
• Implement a Kafka consumer that processes messages exactly-once by starting a workflow per event with idempotency guarantees.
• Schedule recurring workflows using cron syntax and durably sleep workflows for days or weeks, surviving restarts and interruptions.
• Manage AI agent workflows that call unreliable external APIs, with automatic retry and recovery on failure.
• Enqueue background tasks with flow control, timeouts, rate limiting, and deduplication using only Postgres as the backing store.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

DBOS adds durable workflows and queues to Python applications by checkpointing execution state in Postgres, allowing programs to automatically recover from failures without external orchestration infrastructure.

Yes. DBOS is actively maintained, MIT-licensed, has low install friction, and solves a genuine problem—reliable failure handling without external infrastructure—for applications that need durable workflows. It is production-ready (Development Status 5) and supports current Python versions. Install if your application requires fault-tolerant execution and you want to avoid managing a separate orchestration service.

Install

dbos on PyPI

pip install dbos

uv add dbos

poetry add dbos

Installing dbos

Before you install

Low install friction with a pure Python wheel and six standard runtime dependencies. Active maintenance with a recent release 15 days ago. Supports current Python versions (3.10–3.14).

License in practice

MIT license permits unrestricted use, modification, and distribution in both open-source and commercial projects with minimal obligations.

Quickstart

pip install dbos

from dbos import DBOS

@DBOS.step()
def my_step():
    return "done"

@DBOS.workflow()
def my_workflow():
    return my_step()

DBOS.start_workflow(my_workflow)

Requires a Postgres database connection; DBOS must be initialized with a database_url before workflows can run.

Verify before relying

• Whether the package includes built-in support for distributed execution across multiple processes or machines.
• Performance characteristics and latency overhead of checkpoint operations for high-throughput workflows.
• Compatibility with async/await patterns beyond the AsyncIO framework classifier.

Package facts

Evidence: dbos-2.29.0-py3-none-any.whl

Tags

More Python Modules packages