What we see (one of the worker processes - selected randomly?) is handling the events feels weird to me. I’ve created a lengthy post about this:
That is a larger problem on its own, this is why I created separate post. I will let Katello devs to investigate and help you in this one.
@Jonathon_Turel remember we have a nice telemetry API in Foreman core which is not utilized at all for Katello events. It is extremely easy to add statements to monitor amount of events, events per minute, length of queues and other important metrics which could help.
What would be also possible to call Ruby VM stat before and after each event is processed and make a reading of amount of allocated objects. This can be reported per event and easily aggregated later.
Here is a reminder to use our telemetry API for such information: Foreman telemetry API for developers
Yes, title set via setproctitle are only read by some tools like ps or top.