Background task queue for Python backed by Redis, a minimal Celery.
Project description
WakaQ
Background task queue for Python backed by Redis, a super minimal Celery. Read about the motivation behind this project on this blog post and the accompanying Hacker News discussion. WakaQ is currently used in production at WakaTime.com. WakaQ is also available in TypeScript.
Features
- Queue priority
- Delayed tasks (run tasks after a timedelta eta)
- Scheduled periodic tasks
- Broadcast a task to all workers
- Task soft and hard timeout limits
- Optionally retry tasks on soft timeout
- Combat memory leaks with
max_mem_percent
ormax_tasks_per_worker
- Super minimal
Want more features like rate limiting, task deduplication, etc? Too bad, feature PRs are not accepted. Maximal features belong in your app’s worker tasks.
Installing
pip install wakaq
Using
import logging
from datetime import timedelta
from wakaq import WakaQ, Queue, CronTask
# use constants to prevent misspelling queue names
Q_HIGH = 'a-high-priority-queue'
Q_MED = 'a-medium-priority-queue'
Q_LOW = 'a-low-priority-queue'
Q_OTHER = 'another-queue'
Q_DEFAULT = 'default-lowest-priority-queue'
wakaq = WakaQ(
# List your queues and their priorities.
# Queues can be defined as Queue instances, tuples, or just a str.
queues=[
(0, Q_HIGH),
(1, Q_MED),
(2, Q_LOW),
Queue(Q_OTHER, priority=3, max_retries=5, soft_timeout=300, hard_timeout=360),
Q_DEFAULT,
],
# Number of worker processes. Must be an int or str which evaluates to an
# int. The variable "cores" is replaced with the number of processors on
# the current machine.
concurrency="cores*4",
# Raise SoftTimeout in a task if it runs longer than 30 seconds. Can also be set per
# task or queue. If no soft timeout set, tasks can run forever.
soft_timeout=30, # seconds
# SIGKILL a task if it runs longer than 1 minute. Can be set per task or queue.
hard_timeout=timedelta(minutes=1),
# If the task soft timeouts, retry up to 3 times. Max retries comes first
# from the task decorator if set, next from the Queue's max_retries,
# lastly from the option below. If No max_retries is found, the task
# is not retried on a soft timeout.
max_retries=3,
# Combat memory leaks by reloading a worker (the one using the most RAM),
# when the total machine RAM usage is at or greater than 98%.
max_mem_percent=98,
# Combat memory leaks by reloading a worker after it's processed 5000 tasks.
max_tasks_per_worker=5000,
# Schedule two tasks, the first runs every minute, the second once every ten minutes.
# Scheduled tasks can be passed as CronTask instances or tuples. To run scheduled
# tasks you must keep a wakaq scheduler running as a daemon.
schedules=[
# Runs mytask on the queue with priority 1.
CronTask('* * * * *', 'mytask', queue=Q_MED, args=[2, 2], kwargs={}),
# Runs mytask once every 5 minutes.
('*/5 * * * *', 'mytask', [1, 1], {}),
# Runs anothertask on the default lowest priority queue.
('*/10 * * * *', 'anothertask'),
],
# Run tasks in foreground without workers, only for testing.
synchronous_mode=False,
)
# timeouts can be customized per task with a timedelta or integer seconds
@wakaq.task(queue=Q_MED, max_retries=7, soft_timeout=420, hard_timeout=480)
def mytask(x, y):
print(x + y)
@wakaq.task
def anothertask():
print("hello world")
@wakaq.wrap_tasks_with
def custom_task_decorator(fn):
def inner(*args, **kwargs):
# do something before each task runs
fn(*args, **kwargs)
# do something after each task runs
return inner
if __name__ == '__main__':
# add 1 plus 1 on a worker somewhere
mytask.delay(1, 1)
# add 1 plus 1 on a worker somewhere, overwriting the task's queue from medium to high
mytask.delay(1, 1, queue=Q_HIGH)
# print hello world on a worker somewhere, running on the default lowest priority queue
anothertask.delay()
# print hello world on a worker somewhere, after 10 seconds from now
anothertask.delay(eta=timedelta(seconds=10))
Deploying
Optimizing
See the WakaQ init params for a full list of options, like Redis host and Redis socket timeout values.
When using in production, make sure to increase the max open ports allowed for your Redis server process.
When using eta tasks a Redis sorted set is used, so eta tasks are automatically deduped based on task name, args, and kwargs.
If you want multiple pending eta tasks with the same arguments, just add a throwaway random string to the task’s kwargs for ex: str(uuid.uuid1())
.
Running as a Daemon
Here’s an example systemd config to run wakaq-worker
as a daemon:
[Unit]
Description=WakaQ Worker Service
[Service]
WorkingDirectory=/opt/yourapp
ExecStart=/opt/yourapp/venv/bin/python /opt/yourapp/venv/bin/wakaq-worker --app=yourapp.wakaq
RemainAfterExit=no
Restart=always
RestartSec=30s
KillSignal=SIGINT
LimitNOFILE=99999
[Install]
WantedBy=multi-user.target
Create a file at /etc/systemd/system/wakaqworker.service
with the above contents, then run:
systemctl daemon-reload && systemctl enable wakaqworker
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file WakaQ-2.1.25.tar.gz
.
File metadata
- Download URL: WakaQ-2.1.25.tar.gz
- Upload date:
- Size: 19.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/3.7.1 importlib_metadata/4.8.2 pkginfo/1.8.2 requests/2.27.1 requests-toolbelt/0.9.1 tqdm/4.62.3 CPython/3.9.18
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 44952a982a1e61b5c7620d06f0c694c916d91d3c7b5a3ae139e958bcc41881af |
|
MD5 | fc20836528a63e1a4dca74d2db8f14ac |
|
BLAKE2b-256 | 32c6e6146d23cd28ea8571b62142e802cceceeb9327a4de78ce78552e71d1086 |