You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Robert Metzger (JIRA)" <ji...@apache.org> on 2019/04/29 13:01:00 UTC
[jira] [Commented] (FLINK-12202) Consider introducing batch metric
register in NetworkEnviroment
[ https://issues.apache.org/jira/browse/FLINK-12202?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16829238#comment-16829238 ]
Robert Metzger commented on FLINK-12202:
----------------------------------------
Hey [~azagrebin], thanks a lot for opening a ticket in the FLINK bugtracker. I just manually assigned the ticket to a component.
For future tickets, please remember to always assign a new issue to a component, so that the component owner can pick it up.
> Consider introducing batch metric register in NetworkEnviroment
> ---------------------------------------------------------------
>
> Key: FLINK-12202
> URL: https://issues.apache.org/jira/browse/FLINK-12202
> Project: Flink
> Issue Type: Sub-task
> Components: Runtime / Network
> Reporter: Andrey Zagrebin
> Assignee: zhijiang
> Priority: Major
>
> As we have some network specific metrics registered in TaskIOMetricGroup (In/OutputBuffersGauge, In/OutputBufferPoolUsageGauge), we can introduce batch metric registering in NetworkEnviroment.registerMetrics(ProxyMetricGroup, partitions, gates), where task passes its TaskIOMetricGroup into ProxyMetricGroup. This way we could break a tie between task and NetworkEnviroment. TaskIOMetricGroup.initializeBufferMetrics, In/OutputBuffersGauge, In/OutputBufferPoolUsageGauge could be moved into NetworkEnviroment.registerMetrics and network code.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)