Dear MLServer team,
MLServer's Kafka integration currently works only with the PLAINTEXT protocol. For secure deployments, the ability to connect to Kafka over SSL/TLS is required.
Is it possible to extend the Kafka settings to accept SSL configuration fields ? It seems that they could be passed in the settings.json file and/or the environment variables prefixed with MLSERVER_. These settings could be passed via the Settings pydantic model to the MServer class to be consumed by the aiokafka API when instantiating the consumer and producer with SSL context.
Best regards.