[
https://issues.apache.org/jira/browse/FLINK-24823?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
douyongpeng closed FLINK-24823.
-------------------------------
Resolution: Fixed
> Yarn-session report metrics error
> ---------------------------------
>
> Key: FLINK-24823
> URL: https://issues.apache.org/jira/browse/FLINK-24823
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Metrics
> Affects Versions: 1.11.2
> Environment: Flink 1.11
>
> Reporter: douyongpeng
> Priority: Minor
>
> I run 2 streaming jobs in yarn-session。
> I report Metrics to graphite_exporter and something was wrong。
> {code:java}
> * collected metric "flink_jobmanager_fullRestarts" { label:<name:"host"
> value:"xx-xx-xxx-xxx" > label:<name:"job_name" value:"flow-session" >
> gauge:<value:0 > } was c
> ollected before with the same name and label values{code}
> I think some default metric was conflict.
> I use this to know how many jobs is running. This error make graphite crash.
> Is this a bug?
> yarn-session start shell
>
> {code:java}
> exec yarn-session.sh -nm flow-session -D
> metrics.reporter.grph.prefix="flink.flow-session" -D
> env.java.opts="-Djob_name=flow-session" -d {code}
> flink-conf
> {code:java}
> metrics.reporter.grph.class:
> org.apache.flink.metrics.graphite.GraphiteReporter
> metrics.reporter.grph.host: xx-xx-xxx-xxx
> metrics.reporter.grph.port: 9109
> metrics.reporter.grph.protocol: TCP{code}
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
