Active Job reporter reports incorrect data if a job is run within another job via perform_now

[jonleighton]

Issue Description

Sentry's Active Job reporting code works by prepending a module to overload the perform_now method.

This has the outcome that if one job runs the code in another job via perform_now, and an error occurs, the transaction and context will be reported as the inner job, even though it's the outer job being run.

Similarly, we may not even be in a job run at all, e.g. if some controller calls perform_now on a job, the context will be wrong. It will look like the error occurred in a worker process, when it actually occurred in a web process.

Reproduction Steps

class AJob < ApplicationJob
  def perform
    BJob.perform_now
  end
end

class BJob < ApplicationJob
  def perform
    raise "omg"
  end
end

Call AJob.perform_later.

Expected Behavior

Transaction/context is reported from AJob.

Actual Behavior

Transaction/context is reported from BJob.

Ruby Version

3.3.5

SDK Version

5.21.0

Integration and Its Version

No response

Sentry Config

No response

[sl0thentr0py]

right, technically if we're in the same process and not in the worker, we could just start a span instead. I'll have to see what the best way of detecting this is though, or maybe not patch perform_now but the enqueue/execute separately.