execute code with workhorse PID after the workhorse exits

[backwardspy]

i am using rq together with prometheus client for metrics exposition. i have the prometheus client in multiprocess mode and i need to call a function (multiprocess.mark_process_dead(pid)) to let it know when a fork has completed so it can perform some necessary cleanup related to the given PID.

my first thought was a custom worker class, but i'm having trouble finding a convenient place to hook into the worker cleanup process that has access to the child's PID.

this is where i'm at now:

class CustomWorker(Worker):
    def execute_job(self, job, queue):
        super().execute_job(job, queue)
        mark_process_dead(self.horse_pid)

but i think that's too late in the process as self.horse_pid is already set back to 0 by that point.

i appreciate any advice you can offer, and thanks for the awesome project!

[selwin]

Hey, sorry for the late reply. You're right, I don't see an obvious place to do this. There are a few places where horse_pid is set to 0 like here https://github.com/rq/rq/blob/master/rq/worker.py#L1335 but this would only work for successful job executions.

I think we'd need to modify the code base a little bit to store worker and horse information in Result object such that it's easily accessible.