Skip to content

Latest commit

 

History

History
73 lines (55 loc) · 2.33 KB

README.md

File metadata and controls

73 lines (55 loc) · 2.33 KB
Raw

Carrot -- A simple task queue for Django

Why this was made:

  1. It's very easy to use.
  2. It decouples the task code from the task manager code.

Quick Setup:

  1. Install rabbitmq-server (default on localhost)
yum install rabbitmq-server
service rabbitmq-server start
  1. Install carrot
pip install djangocarrot
manage.py makemigrations carrot
manage.py migrate carrot
  1. Add carrot to installed apps (settings.py) INSTALLED_APPS = ['carrot']

Basic Usage:

from carrot.models import Task  

Task.objects.create(kallable="myapp.mymodule.mykallable")

Task arguments:

  • kallable (required): location to callable, e.g. animals.Dog.bark
  • args (optional): list of positional arguments to pass to the callable
  • kwargs (optional): map of key/value pairs to pass to the callable
  • worker (optional): request a specific worker to execute task

Consume tasks

The tasks need to be consumed. This can easily be accomplished with the WorkerService

from carrot.utils import WorkerService
WorkerService().start()

The worker service can also be daemonized with a management command manage.py carrot-workers

Workers (Advanced Usage)

A default worker is already available to use for tasks and this step is not necessary. However, new workers may be created to handle specific tasks. For example, let's say you need a worker which only processes one task at a time so as not to overwhelm some system resource.

from carrot.models import Worker, Task

worker = Worker(name='fileprocesser')
worker.description = 'Used for processing large files one at a time'
worker.concurrency = 1
worker.save()

Task.objects.create(kallable='example.processFile', args=['/tmp/huge1.wav'], kwargs={'bitrate': 1411}, worker=worker)
Task.objects.create(kallable='example.processFile', args=['/tmp/huge2.wav'], kwargs={'bitrate': 1411}, worker=worker)
Task.objects.create(kallable='example.processFile', args=['/tmp/huge3.wav'], kwargs={'bitrate': 1411}, worker=worker)

Worker arguments:

  • name (required): an arbitrary name to identify the worker
  • description (optional): a description of worker and it's duties
  • concurrency (optional) (default -> 4): how many tasks this worker can process concurrently
  • enabled (optional) (default: True): whether or not this worker should be processing tasks