Tom Bentley

### Replication factor change is a special-case of partition reassignment As mentioned, Kafka doesn't support Replication Factor (RF) as a first class concept. It's just a convenience at the time...

> OT, but how do you model this in the `api`? > > ```yaml > throttledRates: > 1: # broker 1 > leader: 70000000 > follower: 70000000 > 2: #...

> > OT, but how do you model this in the `api`? > > ```yaml > > throttledRates: > > 1: # broker 1 > > leader: 70000000 > >...

> > I don't think pre-3.11 will prevent you from having a status object in your CR, it just won't be seen as a status subresource. I will experiment with...

Oh, about the fabric8 support, see https://github.com/fabric8io/kubernetes-client/issues/417

> Further complication comes if we want to make use of subresource support when it's available. So the question there is: How can we discover when the k8s supports customresource...

### Constraints for decreasing the replication factor In most-important first order: 1. The preferred leader should not be changed. 2. The retained replicas should be from different racks (where possible)...

@fibersel you're welcome to have a look into it. There are at least a couple of other people who are also interested. I'll tell you what I've told them: You're...

I think since I originally made this comment we've change the `DefaultAdminClientProvider` to using PEM certificates. But the basic idea still holds. Here are the kind of abstractions I would...

I think this kind of race will become a lot more easy to detect once [Kafka's support for topic ids](https://issues.apache.org/jira/browse/KAFKA-8872) matures and topic ids can be accessed via the Admin...