Details
-
Sub-task
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
I think we should consider collapsing multiple nodeAdded / nodeLost events, ie. generate a single event with information about multiple node names, if there were multiple changes to the cluster in a short time.
Processing a single event that records N changes could help us to react faster to cluster changes than processing N events each with a single change, and with waitFor in between.