Revoking Tasks in Celery
Sometimes you need to cancel a task that has already been submitted or is currently executing. Celery provides the revoke command for this purpose.
1. Getting the Task ID
To revoke a task, you need its ID. This ID is returned when you call the task.
from tasks import long_running_task
# Start the task
result = long_running_task.delay(10)
task_id = result.id
print(f"Task started with ID: {task_id}")2. Revoking a Pending Task
If the task is still in the queue (waiting for a worker), you can revoke it. When a worker picks it up, it will see the revocation flag and discard the task without executing it.
from tasks import app
# Revoke using the Celery app instance
app.control.revoke(task_id)3. Revoking a Running Task (Termination)
If the task is already executing inside a worker, a standard revoke will not stop it. You must set terminate=True.
app.control.revoke(task_id, terminate=True)Specifying Signals
By default, terminate=True sends the SIGTERM signal. You can specify a different signal, such as SIGKILL, if the task refuses to stop.
import signal
app.control.revoke(task_id, terminate=True, signal=signal.SIGKILL)Warning: Terminating a task mid-execution can leave your data in an inconsistent state or leave resources (like file handles) open. Use with caution.
4. Persistent Revocations
By default, the list of revoked tasks is stored in the memory of the workers. If a worker restarts, it forgets which tasks were revoked.
To make revocations persist across restarts, you must configure a persistent state for the workers.
Start the worker with the -S (statedb) flag:
celery -A tasks worker -S /var/run/celery/worker.state