flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From GitBox <...@apache.org>
Subject [GitHub] Tom-Goong commented on a change in pull request #7820: [FLINK-11742][Metrics]Push metrics to Pushgateway without "instance"
Date Fri, 01 Mar 2019 04:34:05 GMT
Tom-Goong commented on a change in pull request #7820: [FLINK-11742][Metrics]Push metrics to
Pushgateway without "instance"
URL: https://github.com/apache/flink/pull/7820#discussion_r261476961
 
 

 ##########
 File path: flink-metrics/flink-metrics-prometheus/src/main/java/org/apache/flink/metrics/prometheus/PrometheusPushGatewayReporter.java
 ##########
 @@ -73,7 +77,7 @@ public void open(MetricConfig config) {
 	@Override
 	public void report() {
 		try {
-			pushGateway.push(CollectorRegistry.defaultRegistry, jobName);
+			pushGateway.push(CollectorRegistry.defaultRegistry, jobName, instance);
 
 Review comment:
   First of all, thank you very much for the information provided, which will help me better
understand Flink related knowledge.
   
   This question depends on how we view the Flink cluster.
   
   First, consider the cluster as a black box. JM is the same as "dispatch" in SpringMVC.
That whole cluster is a Promethues Job.
   
   Second, the JM class is compared to a more feature-rich nginx. Then different jobs correspond
to different Promethues Jobs.
   
   Third, treat each Flink Job as a microservice cluster. The same function of Task and even
Sub-Task is a specific micro-service function, a series of associated micro-services complete
the entire business. Because the above two methods need to solve a problem, different TM runs
this different Flink Job's task, the same task runs on different TMs. This means that the
TM's logo cannot accurately separate the tasks of different Flink Jobs, metrics of the same
name still conflict . JM and TM are just the running containers for jobs. So in theory, we
need to be precise to the Task and even the Sub-Task level, then use the Job prefix to group
them.
   
   To solve this problem perfectly, we need a comprehensive understanding of the entire Metric
system. Maybe we should close this PR and decide how to deal with it after a thorough discussion
in the community.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services

Mime
View raw message