http://jira.hyperic.com/browse/HHQ-3715
Currently the agent scheduler in HQ is limited to a single thread which is responsible for sending out schedules. This was done to "throttle" the number of schedules sent out, but does have a couple of undesirable side-effects:
1) Any hang in scheduling can cause other metrics to not be scheduled. (Captured in HHQ-3694)
2) In a mass-schedule scenario, it can take minutes to potentially hours for agents to get their schedules
I think it would be a nice improvement if we could configure a thread pool that would be responsible for sending these schedules out. This would alleviate issues seen in #1 above with the added benefit of addressing IO bottlenecks from #2.
Anonymous