# 🚀 PgQueuer: PostgreSQL-backed background job and task queue for Python

[![CI](https://github.com/janbjorge/pgqueuer/actions/workflows/ci.yml/badge.svg)](https://github.com/janbjorge/pgqueuer/actions/workflows/ci.yml?query=branch%3Amain) [![pypi](https://img.shields.io/pypi/v/pgqueuer.svg)](https://pypi.python.org/pypi/pgqueuer) [![downloads](https://static.pepy.tech/badge/pgqueuer/month)](https://pepy.tech/project/pgqueuer) [![stars](https://img.shields.io/github/stars/janbjorge/pgqueuer?style=flat)](https://github.com/janbjorge/pgqueuer/stargazers) [![versions](https://img.shields.io/pypi/pyversions/pgqueuer.svg)](https://github.com/janbjorge/pgqueuer)

[📚 Docs](https://janbjorge.github.io/pgqueuer/) · [💻 Source](https://github.com/janbjorge/pgqueuer/) · [💬 Discord](https://discord.gg/C7YMBzcRMQ)

**Your PostgreSQL database is already a job queue.**

PgQueuer turns PostgreSQL into a fast, reliable background job processor. Jobs live in the same database as your application data. One stack, full ACID guarantees, and no separate message broker to run.

## AI-assisted setup

Already using a coding agent? Copy the onboarding prompt from the
[PgQueuer documentation](https://janbjorge.github.io/pgqueuer/#let-an-agent-introduce-you).
The prompt directs your agent to the maintained setup route and canonical docs.

## Features

- 💡 **Minimal footprint**: one `pip install`; bring your existing PostgreSQL connection and start enqueueing
- 🔁 **Transactional enqueue**: commit a job in the same transaction as your data; no dual-write drift
- ⚛️ **Safe concurrency**: workers claim jobs with `FOR UPDATE SKIP LOCKED` (never double-processed), with per-entrypoint limits and serialized dispatch when you need them
- 🚀 **Instant dispatch**: `LISTEN/NOTIFY` wakes workers the moment a job lands (with a polling fallback)
- ⏰ **Scheduling & deferral**: cron-style recurring tasks and `execute_after`, no separate scheduler process
- 📊 **Observability**: completion tracking, Prometheus metrics, tracing (Logfire/Sentry), and a live dashboard
- 🧪 **In-memory mode**: run the whole queue without Postgres for tests and prototyping

## Why PostgreSQL?

If you already run PostgreSQL, it can do double duty as your job queue. That means one fewer service to operate, and your queue and data stay consistent because they share the same database and transactions.

```text
┌──────────┐  enqueue   ┌────────────┐  NOTIFY   ┌──────────┐
│ Your App │───────────▶│            │──────────▶│ Worker 1 │──┐
└──────────┘            │            │           └──────────┘  │
                        │ PostgreSQL │  NOTIFY   ┌──────────┐  │
                        │            │──────────▶│ Worker 2 │──┤
                        │            │           └──────────┘  │
                        │            │  NOTIFY   ┌──────────┐  │
                        │            │──────────▶│ Worker N │──┤
                        └────────────┘           └──────────┘  │
                              ▲  FOR UPDATE SKIP LOCKED         │
                              └─────────────────────────────────┘
```

## Installation

PgQueuer targets Python 3.11+ and PostgreSQL 14+:

```bash
pip install pgqueuer
pgq install        # create tables and functions in your database
```

The CLI reads `PGHOST`, `PGUSER`, `PGDATABASE` and related environment variables. Use `pgq sql install` to preview the SQL (or pipe it to another client: `pgq sql install | psql`), `--prefix myapp_` to namespace tables, or `pgq uninstall` to remove the schema. Code reads the same names from `PGQUEUER_PREFIX` and related environment variables, or from `settings=DBSettings(prefix="myapp_")` passed to `PgQueuer` and `Queries`.

## Quick Start

PgQueuer pairs **consumers** (workers that process jobs) with **producers** (code that enqueues jobs).

### 1. Define a consumer

Each entrypoint is a job handler. Run it with the CLI: `pgq run examples.consumer:main`.

```python
import asyncpg
from pgqueuer import PgQueuer
from pgqueuer.db import AsyncpgDriver
from pgqueuer.models import Job

async def main() -> PgQueuer:
    connection = await asyncpg.connect()
    pgq = PgQueuer(AsyncpgDriver(connection))

    @pgq.entrypoint("fetch")
    async def process(job: Job) -> None:
        print(f"Processed: {job!r}")

    return pgq
```

### 2. Enqueue jobs

From your web app, script, or anywhere else with a database connection:

```python
import asyncpg
from pgqueuer.db import AsyncpgDriver
from pgqueuer.queries import Queries

async def main() -> None:
    connection = await asyncpg.connect()
    queries = Queries(AsyncpgDriver(connection))
    await queries.enqueue("fetch", b"hello world")
```

The job arrives instantly via `LISTEN/NOTIFY`, and your consumer's `process` handler picks it up.

### Enqueue inside a transaction

This is what a database-backed queue buys you: the job and your business data commit together, or not at all.

```python
order_id = 42

async with connection.transaction():
    await connection.execute(
        "INSERT INTO orders (id, status) VALUES ($1, 'paid')", order_id
    )
    await queries.enqueue("send_receipt", str(order_id).encode())
    # If the transaction rolls back, the job is never enqueued.
```

## Run without a database

`PgQueuer.in_memory()` is a drop-in replacement that implements the same ports as the real backend, so your handlers stay identical. Good for unit tests and prototyping.

```python
import asyncio
from pgqueuer import PgQueuer
from pgqueuer.models import Job
from pgqueuer.domain.types import QueueExecutionMode

async def main() -> None:
    pq = PgQueuer.in_memory()

    @pq.entrypoint("send_email")
    async def send_email(job: Job) -> None:
        print(f"Sending: {job.payload!r}")

    await pq.qm.queries.enqueue(["send_email"], [b"alice"], [0])
    await pq.qm.run(mode=QueueExecutionMode.drain)

asyncio.run(main())
```

The in-memory adapter has no durability or multi-process coordination, so use the PostgreSQL backend for production. See the [in-memory reference](https://janbjorge.github.io/pgqueuer/reference/in-memory/).

## Documentation

| Topic | What's inside |
|-------|---------------|
| [Core concepts](docs/getting-started/core-concepts.md) | Consumers, producers, entrypoints, the job lifecycle |
| [Scheduling](docs/guides/scheduling.md) | Cron-style recurring tasks and deferred execution |
| [Concurrency control](docs/guides/concurrency-control.md) | Per-entrypoint limits and serialized dispatch |
| [Completion tracking](docs/guides/completion-tracking.md) | Wait for jobs to finish with `CompletionWatcher` |
| [Shared resources](docs/guides/shared-resources.md) | Inject DB pools, HTTP clients, and models into handlers |
| [Custom executors](docs/guides/custom-executors.md) | Retry strategies and exponential backoff |
| [Drivers](docs/reference/drivers.md) | asyncpg, psycopg async/sync: choosing and configuring |
| [Architecture](docs/reference/architecture.md) | Ports & adapters, `SKIP LOCKED`, design decisions |
| [Observability](docs/integrations/prometheus.md) | Prometheus metrics, [tracing](docs/integrations/tracing.md), and the dashboard |
| [Framework integration](examples/) | FastAPI ([example](examples/fastapi_usage.py)) and Flask ([example](examples/flask_sync_usage.py)) |

## Monitor your queues

Launch the interactive dashboard to watch queue activity in real time:

```bash
pgq dashboard --interval 10 --limit 25
```

```text
+---------------------------+-------+------------+--------------------------+------------+----------+
|          Created          | Count | Entrypoint | Time in Queue (HH:MM:SS) |   Status   | Priority |
+---------------------------+-------+------------+--------------------------+------------+----------+
| 2024-05-05 16:44:26+00:00 |  49   |    sync    |         0:00:01          | successful |    0     |
| 2024-05-05 16:44:27+00:00 |  12   |   fetch    |         0:00:03          | queued     |    0     |
| 2024-05-05 16:44:28+00:00 |   3   |  api_call  |         0:00:00          | picked     |    5     |
+---------------------------+-------+------------+--------------------------+------------+----------+
```

## Development

PgQueuer uses [Testcontainers](https://testcontainers.com/?language=python) to spin up an ephemeral PostgreSQL instance for the test suite. Just have Docker running.

```bash
uv sync --frozen  # install dependencies
uv run ruff check . && uv run ruff format . --check && uv run lint-imports && uv run mypy . && uv run pytest  # full check suite
```

## License

PgQueuer is MIT licensed. See [LICENSE](LICENSE) for details.