Status

Current state: Under Discussion

Discussion thread: here [Change the link from the KIP proposal email archive to your own email thread]

JIRA: KAFKA-6788 [Change the link from KAFKA-1 to your own ticket]

Motivation

In KafkaAdminClient, for some requests like describeGroup and deleteGroup, we will first try to get the coordinator for each requested group id, and then send the corresponding request for that group id. However, different group ids could be hosted on the same coordinator, and these requests do support multi group ids be sent within the same request. So we can consider optimize it by grouping the requests per coordinator destination, in order to do so, we want findCoordinator to batch up the groupID that has the same coordinators.

Now, the existing mechanism is that for an individual group, we would send out a request when we call findCoordinator to find the coordinator's ID.

After we know the group's coordinator, then we will send out another request immediately by calling describeConsumerGroups.

runnable.call(new Call("findCoordinator", deadline, new LeastLoadedNodeProvider()){
	...
	runnable.call(new Call("describeConsumerGroups", deadline, new ConstantNodeIdProvider(nodeId)){...}
	...
}

Notice that for each groupID, at least two requests will be sent. That's why we want to batch up the result from findCoordinator and send to describeConsumerGroups in batches to increase efficiency. 

Proposed Changes

Before

After

Compatibility, Deprecation, and Migration Plan

• What impact (if any) will there be on existing users?
• If we are changing behavior how will we phase out the older behavior?
• If we need special migration tools, describe them here.
• When will we remove the existing behavior?

Rejected Alternatives

If there are alternative ways of accomplishing the same thing, what were they? The purpose of this section is to motivate why the design is the way it is and not some other way.