Github user revans2 commented on the issue:
https://github.com/apache/storm/pull/2203
@HeartSaVioR as far as task ids vs executor ids. There are different ways
of doing this for each of the two metrics systems. For the built in zookeeper
stats they are for the range `[start-task-id, end-task-id]`
https://github.com/apache/storm/blob/466a7ad74da27c1250eedf412a487db409e42c19/storm-client/src/storm.thrift#L540
For the current metrics system they are for a single task, not an executor.
https://github.com/apache/storm/blob/466a7ad74da27c1250eedf412a487db409e42c19/storm-client/src/jvm/org/apache/storm/metric/api/IMetricsConsumer.java#L40
So this means that for things like queues that are specific to an executor,
and not a task, the metrics may be odd in that the first task in an executor
would get all of the metrics and the others would get little or nothing.
I agree that we probably want to have them be by component id and not
executor id. We can put them back together into executors when we query the
data.
---
