[
https://issues.apache.org/jira/browse/FLINK-29970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17634355#comment-17634355
]
Weihua Hu commented on FLINK-29970:
-----------------------------------
[~zhongyangyang], Hi, there are some typo in your configuration. such as:
'metric.reporter.prom.class' should be 'metrics.reporter.prom.class' and
'annotaions' shouldbe 'annotations'.
you can try with these configs:
{code:java}
metrics.reporter.prom.factory.class:
org.apache.flink.metrics.prometheus.PrometheusReporterFactory
metrics.reporters: prom
kubernetes.jobmanager.annotations:
prometheus.io/port:9249,prometheus.io/scrape:true
kubernetes.taskmanager.annotations:
prometheus.io/scrape:true,prometheus.io/port:9249
{code}
And why do you have to use the service to expose port 9249? Maybe You can use
the role: pod to discover flink jobmanager and taskmanager.
> Prometheus cannot collect flink metrics
> ---------------------------------------
>
> Key: FLINK-29970
> URL: https://issues.apache.org/jira/browse/FLINK-29970
> Project: Flink
> Issue Type: New Feature
> Components: Deployment / Kubernetes
> Affects Versions: 1.14.6, 1.15.3
> Reporter: 钟洋洋
> Priority: Minor
>
> When I use the native k8s method to deploy my flink application cluster, if I
> do not manually deploy a Service to expose the 9249 port of my jobmanager and
> taskmanager, the prometheus I deployed in k8s cannot collect the metrics in
> my flink. Should flink generate these services itself when deploying？Or am I
> using it incorrectly?Thanks!
> My deploy command is（Some content is omitted）
> {code:sh}
> flink run-application --target-application \n
> -Dmetrics.reporters=prom \n
> -Dmetric.reporter.prom.class=org.apache.flink.metrics.prometheus.PrometheusReport
> \n
> -Dkubernetes.jobmanager.annotaions=prometheus.io/scrape:true,prometheus.io/port:9249
> \n
> -Dkubernetes.jobmanager.annotaions=prometheus.io/scrape:true,prometheus.io/port:9249
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
