Are there more documentation on the inner workings of the following? I have
read the intro's and things are still not that clear to me:
- how nimbus, worker processes and supervisors work? how/what they interact
about with ZooKeeper? How do they find out if a spout is not responding?
- DRPC server? What guarantees the availability of this one? is this also
managed by the supervisors/nimbus? Or are we on our own on this one?
- Overall is the Request/Response is an appropriate model for Storm? If so why
can't a spout be used in placed of a DRPC server?
- How is the completion of a bolt task is determined? Especially if grouping is
involved?
