saq 0.1.1

Distributed Python job queue with asyncio and redis

SAQ

SAQ (Simple Async Queue) is a simple and performant job queueing framework built on top of asyncio and redis. It is used for processing background jobs with workers. For example, you could use SAQ to schedule emails, run long running queries, or data analysis.

It uses aioredis >= 2.0.

It is similar to RQ and heavily inspired by ARQ. Unlike RQ, it is async and thus significantly faster if your jobs are async. Even if they are not, SAQ is still considerably faster due to lower overhead.

SAQ optionally comes with a simple UI for monitor workers and jobs.

Install

# minimal install
pip install saq

# web + hiredis
pip install saq[web,hiredis]

Usage

usage: saq [-h] [--workers WORKERS] [--verbose] [--web] settings

Start Simple Async Queue Worker

positional arguments:
  settings           Namespaced variable containing worker settings eg: eg module_a.settings

options:
  -h, --help         show this help message and exit
  --workers WORKERS  Number of worker processes
  --verbose, -v      Logging level: 0: ERROR, 1: INFO, 2: DEBUG
  --web              Start web app

Example

import asyncio

from saq import Queue

# all functions take in context dict and kwargs
async def test(ctx, *, a):
    await asyncio.sleep(0.5)
    # result should be json serializable
    # custom serializers and deserializers can be used through Queue(dump=,load=)
    return {"x": a}

async def startup(ctx):
    await ctx["db"] = create_db()

async def shutdown(ctx):
    await ctx["db"].disconnect()

async def before_process(ctx):
    print(ctx["job"], ctx["db"])

async def after_process(ctx):
    pass

queue = Queue.from_url("redis://localhost")

settings = {
    "queue": queue,
    "functions": [test],
    "concurrency": 10,
    "startup": startup,
    "shutdown": shutdown,
    "before_process": before_process,
    "after_process": after_process,
}

To start the worker, assuming the previous is available in the python path

saq module.file.settings

To enqueue jobs

# schedule a job normally
job = await queue.enqueue("test", a=1)

# wait 1 second for the job to complete
await job.refresh(1)
print(job.results)

# schedule a job in 10 seconds
await queue.enqueue("test", a=1, scheduled=time.time() + 10)

Demo

Start the worker

saq examples.simple.settings --web

Navigate to the web ui

Enqueue jobs

python examples/simple.py

Comparison to ARQ

SAQ is heavily inspired by ARQ but has several enhancements.

1. Avoids polling by leveraging BLMOVE or RPOPLPUSH and NOTIFY
   1. SAQ has much lower latency than ARQ, with delays of < 5ms. ARQ's default polling frequency is 0.5 seconds
   2. SAQ is up to 8x faster than ARQ
2. Web interface for monitoring queues and workers
3. Heartbeat monitor for abandoned jobs
4. More robust failure handling
   1. Storage of stack traces
   2. Sweeping stuck jobs
   3. Handling of cancelled jobs different from failed jobs (machine redeployments)
5. Before and after job hooks
6. Easily run multiple workers to leverage more cores

Development

python -m venv env
source env/bin/activate
pip install -e .[dev,web]
docker run -p 6379:6379 redis
./run_checks.sh