Hi everyone,
Linkis is a distributed microservices system, including multi long-running and dynamic / ac-hoc (the engine conn) instances. It might be better if we can monitor each JVM service, rather than the fixed host server. The current pain point is how to enable monitoring system connect to each instance and retrieve the metrics in a standard way. The Prometheus can help us solve this problem, by leveraging the feature of service discovery (SD). The current register center used in Linkis is Eureka, and it's support by prometheus as one of the available SD configuration, whcih allows retrieving scrape targets using the Eureka REST API. And Prometheus will periodically check the REST endpoint and create a target for every app instance. Based on this, we can enable Linkis to provide the scrape targets in Eureka metadata, and open the metrics endpoint for each instance. Once the instances can be monitored in Prometheus, we can setup the alter channel in Prometheus AlertManager and dashboard in Grafana. The overall monitoring process can be designed in the following way: The feature and corresponding use cases can be referrf in [Feature] Monitor Linkis based on Prometheus #1656 Welcome the suggestion or idea from you, to make it better. Thanks! Sun Shun
