[ https://issues.apache.org/jira/browse/FLINK-28816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17575798#comment-17575798 ]
Yang Wang commented on FLINK-28816: ----------------------------------- TBH, I hesitate to add pod related metrics in the flink-kubernetes-operator. In my opinion, many companies have already build their own monitoring system based on the promethus, which should include the pod creation metrics. The monitoring is not only for Flink workloads, but also the online systems, spark workloads, etc. Moreover, I do not think the operator has enough information to calculate the pod creation cost/failure-rate metrics except for querying from K8s. > Include some metrics for the pod created in operator > ---------------------------------------------------- > > Key: FLINK-28816 > URL: https://issues.apache.org/jira/browse/FLINK-28816 > Project: Flink > Issue Type: Improvement > Components: Kubernetes Operator > Reporter: Aitozi > Priority: Major > > Currently, the metrics are around the operator self operation. In our use > case, we also want to measure the metric especially about the flink pod's > create time cost, pod create failure rate metrics, I think the operator is > the best place to put/collect these metrics. > -- This message was sent by Atlassian Jira (v8.20.10#820010)