[Discussion] Rebalance command behavior is not intuitive.

[Satish Duggana <sd...@hortonworks.com>]

Hi,
Rebalance command gives different options to rebalance a topology like below.

storm rebalance topology-name [-w wait-time-secs] [-n new-num-workers] [-e component=parallelism]*

You can say "storm rebalance help” to get help note on the command.

Whenever –n option is used, it always changes no of workers. It seems to take parallelism as a hint and it may or may not rebalance topology with the given parallelisms. When multiple components are given with different parallelism, some components are updated with given parallelism and some other components were ignored. This is confusing as user can not really know whether rebalance command ignored given parallelism or there may be some other issue while it tried to rebalance.

What is the actual intended behavior of "storm rebalance”?
Enhancement may be needed for rebalance to return the result of whether it has ignored or applied the given configuration.

For ex:
storm rebalance "word-count-1"  -n 3 -e “count=7"
workers - 3
Executors:
_ackers - 3
spout - 5
split - 4
count - 6 (remained same, no change in executors)

storm rebalance "word-count-1" -e “spout=5” “count=8"
workers - 3
Executors:
_ackers - 3
spout - 5  (remained same, no change in executors)
split - 4
count - 8 (count is updated)


Please let me know your comments.

Thanks,
Satish.