Hi all, Here is two suggestions for the Kafka Connect API that I prefer to present you first before creating KIPs.
1/ Currently, in Kafka Connect we distinguish two configuration types: the worgker config and the connector config. The connector config is accessible from the Connector class through th method "*void start(Map<String, String> props)*" but developers can't access the originals worker configuration (i.e bootstrap.servers, etc). I think some use cases could require an internal producer or consumer created from the Connector/Task instance. For example in order to : - To consume a topic for special events which will trigger tasks reconfiguration - To produce messages in order to track tasks progression (the messages could be consumed by a KafkaStreams app in order to expose monitoring UI). Currently, the only way to do this is to define the kafka configuration in each configuration but this is not really developer friendly. 2/ Secondly, it could be useful to have a method returning the connector name accessible from the Connector class. For example, the connector name could be used as an identifier for retrieving sensitives settings from a configuration management server (for example Consul). Do you think this two minor suggestions can deserved KIPs ? Thanks very much. Florian. -- Florian HUSSONNOIS
