On Tuesday 25 Feb 2014 10:17:25 you wrote:
> - Celery went down for a while.
>
> - Celerybeat jobs accumulated in RabbitMQ.
That's weird, the beat stuff is integrated into the celeryd isn't it?
> It's a separate bug, but we should ensure that Celerybeat jobs have a
> very short lifetime. Maybe RabbitMQ or Celery have a mechanism for
> ignoring old jobs. If not, we can add a timestamp check to the job
> implementations themselves.
On Tuesday 25 Feb 2014 10:17:25 you wrote:
> - Celery went down for a while.
>
> - Celerybeat jobs accumulated in RabbitMQ.
That's weird, the beat stuff is integrated into the celeryd isn't it?
> It's a separate bug, but we should ensure that Celerybeat jobs have a
> very short lifetime. Maybe RabbitMQ or Celery have a mechanism for
> ignoring old jobs. If not, we can add a timestamp check to the job
> implementations themselves.
Yep, we need some protection against this.