Ensure jobs run sequentially and in order

In this section, we’ll see how to setup locks. If you want to know more about the locking feature (mainly the why), head to the Discussions section (see About locks).

When deferring a job, we can provide a lock string to the configure method:

my_task.configure(lock=customer.id).defer(a=1)
my_other_task.configure(lock=customer.id).defer(b=2)

Or if we’re deferring the same task with the same lock multiple times, we can call configure just once:

job_description = my_task.configure(lock=customer.id)
job_description.defer(a=1)
job_description.defer(a=2)

In both cases, the second task cannot run before the first one has ended (successfully or not).

Warning

If a task with a scheduled_at argument is deferred with a lock, then following tasks can only run after this one is processed, which may be in a long time.

Similarly, if the oldest task of a lock is in a queue that no worker consumes, the other tasks are blocked.

If you plan to use the same lock for every job deferred from the same task, you can define the value when you register the task:

@app.task(lock="my_lock_value")
def my_task(**kwargs):
    ...

Locks and Priority

When multiple jobs share the same lock, they are processed one at a time in a specific order:

• descending priority (higher priority first)

• ascending creation time (older job first)

If any job with the same lock is running, all other jobs with that lock must wait. A high-priority job cannot jump ahead of a currently running job, regardless of the running job’s priority.