Hello, I’d like to propose a feature enhancement for the kafka-client java library, where implementors can choose a custom HostResolver implementation.
Rationale: we’re currently connecting to a Kafka broker that we don’t control, running in a Windows server environment. This server lists its own machine host name in `advertised.listeners` rather than an IP address. This would normally be fine, except that this name isn’t registered in DNS in our target environment - in fact there are no local DNS servers. Since we intend to replicate this setup in many locations, it becomes cumbersome to also replicate a DNS server just to serve this one name (that will be different in each location). Instead, we plan to use a custom HostResolver implementation that uses a 3rd party library (jcifs-ng) to do the lookup over SMB. Proposal: add a `hostname.resolver` optional configuration property specifying a class to use as the HostResolver implementation. If not specified, it defaults to DefaultHostResolver. This would be similar to other configuration options such as `metric.reporters` and `key.deserializer`. I’ve created a branch that implements this, but there’s no associated JIRA ticket and I don’t have access to create one. Should I do that? Thank you!
