Celery任务多队列注册异常求助（Django1.9+RabbitMQ环境）

Hey there! Let's figure out why your add task is ending up in all four queues (A, B, C, D) instead of just queue A. I've tackled similar Celery routing headaches before, so here are the most common fixes to check step by step:

First off, make sure you're explicitly assigning the queue in the task decorator—this is the most straightforward way to pin a task to a specific queue, and typos here are super easy to miss:

# In your tasks.py file
from celery import app

# Make sure the queue name matches exactly (RabbitMQ is case-sensitive!)
@app.task(queue='A')
def add(x, y):
    return x + y

If you skipped the queue parameter here, Celery will fall back to default routing rules, which might be sending your task to all queues by accident.

Next up, head to your Django settings (or Celery config file) and check these critical settings:

• Avoid wildcard routes targeting all queues: If you have something like this, it will override your task-specific queue assignment:

  # Bad example—this routes EVERY task to all four queues!
  CELERY_ROUTES = {
      '*': {'queue': ['A', 'B', 'C', 'D']},
      # ... other routes
  }
  

• Verify the add task's route entry: If you're using explicit route mappings, ensure it's only pointing to queue A:

  # Correct setup for the add task
  CELERY_ROUTES = {
      'your_app_name.tasks.add': {'queue': 'A'},
      # ... other task routes
  }
  

  Note: Depending on your Celery version, the config key might be task_routes (Celery 4.x+) instead of CELERY_ROUTES.

Sometimes the issue isn't with Celery, but how RabbitMQ is set up:

• If you're using a fanout exchange, any task sent to this exchange will be broadcast to all queues bound to it. Check if queues B, C, D are accidentally bound to the same exchange as queue A.
• Use the RabbitMQ management interface (if you have it enabled) to inspect:
  • Which exchange your add task is using
  • Which queues are bound to that exchange
    Remove any unintended bindings to stop the task from reaching other queues.

Celery can hold onto old configuration settings longer than you'd expect. Try these steps to refresh everything:

• Stop all running Celery workers and beat processes (if you're using beat)
• Clear Django's cache (run python manage.py clearcache if you have the cache framework set up)
• Restart Celery workers with your updated config—make sure you're not passing -Q A,B,C,D to the worker command unless you want it to listen to all queues (this is for worker listening, not task routing, but it's worth checking!)

To narrow down the issue, send a task directly specifying the queue and see if it still lands in all queues:

# In a Django shell or script
from your_app_name.tasks import add
add.apply_async(args=(2, 3), queue='A')

If this test still sends the task to all queues, the problem is definitely in your routing/exchange setup. If it only goes to queue A, then check how you're normally calling the task—you might be accidentally specifying multiple queues in your regular task calls.

内容的提问来源于stack exchange，提问作者SHIVAM JINDAL