redqueue 0.11.0

Redis-backed Python message queue library with List, Streams, delayed tasks, and monitoring.

RedQueue

RedQueue is a Redis-backed Python message queue library with List, Streams, delayed tasks, synchronous APIs, asynchronous APIs, compatibility checks, monitoring hooks, and connection-pool resource management.

Chinese documentation: README-zh-CN.md

Repository: https://github.com/SpringMirror-pear/redqueue.git

Features

• Redis List reliable queue with BLMOVE on Redis >=6.2 and BRPOPLPUSH fallback on older compatible Redis versions.
• Redis Streams backend with consumer groups. Streams require Redis >=5.0.
• Delayed tasks based on Redis Sorted Set.
• Sync client QueueClient and async client AsyncQueueClient.
• Redis connection pool managers for shared sync and async resources.
• Unified exception hierarchy with structured context.
• Monitoring events for publish, consume, ack, nack, retry, dead letter, delay, and backend errors.
• Redis capability detection from INFO server.
• Apache License 2.0.

Compatibility

Runtime:

• Python >=3.9
• redis-py 6.4.0
• Target development environment: Python 3.14.5

Redis:

Feature	Redis requirement	Notes
List blocking consume	>=2.0	Uses BLPOP family compatibility baseline
List reliable move	>=2.2	Uses BRPOPLPUSH; BLMOVE preferred on >=6.2
Streams	>=5.0	Uses XADD, XGROUP CREATE, XREADGROUP
Streams auto claim	>=6.2	Uses XAUTOCLAIM; Redis 5.x uses XPENDING/XCLAIM fallback
Delayed tasks	>=1.2	Uses ZADD and timestamp scores

Installation

pip install redqueue

For local development:

python -m pip install -r requirements.txt

Quick Start

Synchronous List queue:

from redqueue import QueueClient

client = QueueClient.from_url(
    "redis://127.0.0.1:6379/0",
    queue="emails",
    backend="list",
)

message_id = client.publish({"to": "[e-mail]"})
message = client.consume(timeout=1)

if message is not None:
    try:
        print(message.payload)
        client.ack(message)
    except Exception:
        client.retry(message, reason="handler failed")

Streams backend:

from redqueue import QueueClient

client = QueueClient.from_url(
    "redis://127.0.0.1:6379/0",
    queue="events",
    backend="stream",
    consumer_group="redqueue",
    consumer_name="worker-1",
)

client.publish({"event": "created"})
message = client.consume(timeout=1)

Asynchronous client:

import asyncio

from redqueue import AsyncQueueClient


async def main() -> None:
    client = await AsyncQueueClient.from_url(
        "redis://127.0.0.1:6379/0",
        queue="jobs",
        backend="list",
    )
    await client.publish({"task": "sync"})
    message = await client.consume(timeout=1)
    if message is not None:
        await client.ack(message)
    await client.close()


asyncio.run(main())

Delayed task:

from redqueue import QueueClient

client = QueueClient.from_url("redis://127.0.0.1:6379/0", queue="emails")
client.delay({"to": "[e-mail]"}, delay_seconds=60)
released = client.schedule_due(limit=100)

Connection pool management:

from redqueue import QueueClient, RedisConnectionManager

with RedisConnectionManager(
    "redis://127.0.0.1:6379/0",
    max_connections=20,
    health_check_interval=30,
) as manager:
    producer = QueueClient.from_url(
        manager.url,
        queue="emails",
        connection_manager=manager,
    )
    consumer = QueueClient.from_url(
        manager.url,
        queue="emails",
        connection_manager=manager,
    )

    producer.publish({"to": "[e-mail]"})
    message = consumer.consume(timeout=1)
    if message is not None:
        consumer.ack(message)

Async connection pool management:

import asyncio

from redqueue import AsyncQueueClient, AsyncRedisConnectionManager


async def main() -> None:
    async with AsyncRedisConnectionManager(
        "redis://127.0.0.1:6379/0",
        max_connections=20,
    ) as manager:
        client = await AsyncQueueClient.from_url(
            manager.url,
            queue="jobs",
            connection_manager=manager,
        )
        await client.publish({"task": "sync"})


asyncio.run(main())

Branch Model

RedQueue uses a lightweight Git Flow model:

• main: stable release branch. Only release merges and urgent hotfixes land here.
• develop: integration branch for the next minor release.
• feature/<name>: feature work branched from develop, merged back into develop.
• release/<minor>: release stabilization branch, such as release/0.11.
• hotfix/<version>: urgent patch branch from main, merged back to both main and develop.

Documentation

• Chinese README: README-zh-CN.md
• API: docs/API.md
• Examples: examples/README.md
• Changelog: CHANGELOG.md
• Release process: docs/RELEASE.md
• Test guide: tests/README.md
• Contributing: CONTRIBUTING.md
• Code of conduct: CODE_OF_CONDUCT.md

Examples

The examples/ directory contains runnable scripts for synchronous List queues, asynchronous List queues, Streams, delayed tasks, monitoring hooks, custom serializers, and Redis compatibility checks.

PYTHONPATH=src python examples/sync_list_queue.py
PYTHONPATH=src python examples/async_list_queue.py
PYTHONPATH=src python examples/stream_queue.py
PYTHONPATH=src python examples/delayed_tasks.py
PYTHONPATH=src python examples/monitoring_hooks.py
PYTHONPATH=src python examples/custom_serializer.py
PYTHONPATH=src python examples/compatibility_check.py

Testing

PYTHONPATH=src python -m pytest

Run integration tests with a local Redis server:

REDQUEUE_REDIS_URL=redis://127.0.0.1:6379/0 PYTHONPATH=src python -m pytest -m integration

Run availability tests:

PYTHONPATH=src python -m pytest -m availability

Run deterministic in-memory performance tests:

PYTHONPATH=src python -m pytest -m performance

Run real Redis availability tests:

REDQUEUE_REDIS_URL=redis://127.0.0.1:6379/0 PYTHONPATH=src python -m pytest -m "integration and availability"

Run real Redis performance tests:

REDQUEUE_REDIS_URL=redis://127.0.0.1:6379/0 PYTHONPATH=src python -m pytest -m "integration and performance"

Run real Redis concurrency tests:

REDQUEUE_REDIS_URL=redis://127.0.0.1:6379/0 PYTHONPATH=src python -m pytest -m "integration and concurrency"

Availability Results

Latest local run on Python 3.14.5:

• Full test suite without REDQUEUE_REDIS_URL: 77 passed, 8 skipped.
• Real Redis availability suite: 3 passed with redis://127.0.0.1:6379/0.
• Real Redis server: Redis for Windows 5.0.14.1.
• Availability suite: covers List processing recovery, List dead-letter requeue, Streams Redis <5.0 compatibility rejection, Streams Redis 5.x pending recovery fallback, Streams dead-letter requeue, delayed task publish failure rollback, missing delayed payload errors, async List recovery, async Streams dead-letter requeue, and async delayed task rollback.
• Real Redis availability suite additionally validates List recovery, Streams dead-letter requeue, and delayed task rollback against a running Redis server.

Performance Results

The performance suite uses deterministic in-memory Redis fakes. It measures RedQueue overhead without network latency and is intended as a regression baseline, not as a Redis server benchmark.

Real Redis performance tests were also run against Redis for Windows 5.0.14.1. Redis for Windows has extra platform and compatibility overhead, so these numbers are only a local reference and do not represent expected Linux production performance.

Latest local baseline on Python 3.14.5:

Scenario	Operations	Elapsed	Throughput
JSON encode + decode	10,000	0.064742s	154,459 ops/s
Sync List publish + consume + ack	2,000	0.042705s	46,833 ops/s
Sync Streams publish + consume + ack	1,000	0.091318s	10,951 ops/s
Delay schedule + release	1,000	0.033192s	30,127 ops/s
Async List publish + consume + ack	1,000	0.023531s	42,497 ops/s

Latest real Redis baseline on Python 3.14.5 and Redis for Windows 5.0.14.1:

Scenario	Operations	Elapsed	Throughput
Real Redis List publish + consume + ack	200	0.045230s	4,422 ops/s
Real Redis Streams publish + consume + ack	100	0.027108s	3,689 ops/s
Real Redis delay schedule + release	100	0.056498s	1,770 ops/s
Real Redis concurrent List publish + consume + ack	200	1.059509s	189 ops/s

License

Apache License 2.0. See LICENSE.