[GitHub] flink issue #5857: [FLINK-9187][METRICS] add prometheus pushgateway reporter

[lamber-ken]

Github user lamber-ken commented on the issue:

    https://github.com/apache/flink/pull/5857
  
    for example, 
    #### Environment
    - deploy flink job on hadoop yarn using `yarn-cluster` model
    - prometheus collect metrics 
      because we use prometheus in production, so don't want to use graphite or 
ganglia currently etc
    
    #### if not use pushgateway
    because using yarn-cluster model, so we can't know where the jobmanager on 
which machine and 
    we can't to define the prometheus targets.
    
![image](https://user-images.githubusercontent.com/20113411/38981072-1575909c-4384-11e8-8308-9620e06808fb.png)
    
    #### use pushgateway
    
![image](https://user-images.githubusercontent.com/20113411/38981092-1f94e2d0-4384-11e8-998c-902f8951b727.png)
    
    #### prometheus pushgateway
    although pushgateway has some 
[limits](https://prometheus.io/docs/practices/pushing/) currently,
    but it will meet our needs in the future
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    



---