Feature: optionally allow developers to control enqueuing

[ramontayag]

We use sidekiq-cron and noticed that at one point, jobs stop being enqueued. We already use this gem to see if jobs stop running, and thought that we could use the same mechanism to detect times when jobs stop getting enqueued.

Are you open to a feature where we can control the enqueueing of SidekiqAlive::Worker? By default, it would requeue itself, but if you pass requeue: false, it will not requeue itself.

Usage would be like:

SidekiqAlive.start(requeue: false)

requeue defaults to true, and the value is just passed into sa::Server.perform_async(hostname, requeue: requeue) in this line.

Then we enqueue, for every Sidekiq instance and config.time_to_live / 2 every seconds, by calling:

SidekiqAlive::Worker.perform_async(requeue: false)

Does this sound like it'll work? Still need to think through how we will enqueue for each available hostname without hard coding the number of pods we have for Sidekiq.

[arturictus]

Hi @ramontayag,
Sounds interesting been able to modify how jobs are enqueued.
Right now the nice thing about the enqueue is that you do not have to think about it.
what I think could be interesting is to add a config to specify your own worker this way you could
requeue or do whatever you want in this worker.

[ramontayag]

Hello, @arturictus. Yes, it's just a way to turn off the automatic requeueing feature that Sidekiq::Alive has, since we will be the one enqueuing it every N seconds.