If you are using a background queue like Resque, threads are killed after a job completes. In this scenario, I believe errors would fail to send to sentry.

Is there a way to synchronously drain the async error reporting queue? This could be added to a before exit hook on the resque side.