Ensure tasks don’t accumulate in the queue

Some kind of tasks implement actions that absolutely need to be done. Others, for example cleanup tasks, implement actions that would ideally need to be done often, but could easily be missed once in a while. In particular, in case of a slowdown in job processing, your queue can sometimes become filled with jobs that don’t really add much value.

You can tell Procrastinate that a specific set of jobs can never appear more than once in the queue, by using a queueing lock:


Trying to defer a task when the queueing_lock is active will result in an AlreadyEnqueued exception, which you can choose to ignore.

In the command line interface (see Defer a job), you can use --queueing-lock and --ignore-already-enqueued/-i to control how queueing locks are used:

$ procrastinate defer --queueing-lock=maintenance --ignore-already-enqueued \

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

def my_task(**kwargs):

queueing_lock allows a single job in todo status. Meanwhile, it allows multiple jobs to be in doing status.

To enforce that only one job runs at a time while limiting the queue size, you can combine queueing_lock with lock.

You might also want to schedule jobs for such task in the near future, which can be effective to throttle the task.