I have a topology that uses a Kafka spout to read values from a Kafka queue. I used the kafkaSpout that came with storm-0.9.2-incubating.
I observed that the nimbus rebalances the topology very often. The topology suddenly shuts down and starts again with the tasks running on different machines. I wanted to know why storm nimbus is rebalancing my topology, so I observed the storm throughput, latency, load on bolts and even system metrics like cpu and memory utilization, but I could not see a pattern. Can someone explain what are the factors that lead to rebalancing of topology in storm? Thanks, Palak Shah
