the Kafka Exporter deployment is currently using the /metrics endpoints for health checks. This endpoint returns all metrics, so using it for health checks does not seem to be optimal - especially with large clusters (see for example this Slack discussion). The latest version of the Kafka Exporter which we use now seems to have also the /healthz endpoint available which seems to be better for the health checks. It is also used in Kafka Exporter's own Helm Chart (https://github.com/danielqsj/kafka_exporter/blob/master/charts/kafka-exporter/templates/deployment.yaml#L70-L89).