Worker shutdown behaviour

[aminalaee]

Thanks for the library.
This is more of a question but I couldn't find a better place for it.

What happens to the buffered tasks in case of a shutdown (like sending a signal to kill a pod)? Do they go back into the queue?

[clokep]

What happens to the buffered tasks in case of a shutdown (like sending a signal to kill a pod)? Do they go back into the queue?

It somewhat depends on what your shutdown timeout is (time between TERM and KILL signals -- see the docs). If it is longer than flush_interval + the time to process the tasks then you should be OK and the tasks should get processed.

It also will interact with whether you're acking early or late. It really depends on if you want your tasks to execute at most once or at least once.

[aminalaee]

Thanks for the reply. So my only concern for using the library is this and I think you have a solution for that. It'd be great if you could document this.

Thanks again for the great work.