Major Description
We have some basic info about the server available via GET / (currently version information). It'd be nice to have some additional cluster metadata available via the REST API (perhaps under a /cluster endpoint). A cluster ID would be a good starting point, although we'll need to decide whether we really want this as a global view of the cluster or a set of APIs that give you info about the individual worker (where some values should simply be consistent across the cluster).
There are a couple of ways we could implement cluster IDs:
- An entirely new config
- If we could get some unique ID for the Kafka cluster, leverage the name of the config topic. This doesn't require a new worker config, but the name probably isn't ideal – it might include a reasonable prefix, but will also often include the suffix "-config" which will look odd.
- If we could get some unique ID for the Kafka cluster and implement KAFKA-3254, we could automatically generate one as (Kafka cluster ID, topic prefix)
Note that some of these are assuming distributed mode. We'd have to figure out a scheme that can also be applied to standalone clusters. Backwards compatibility is also a concern since we'd rather not introduce any new required configs if possible.
As this is new public API, it'll need a KIP before implementation.