You are viewing a plain text version of this content. The canonical link for it is here.

Posted to jira@kafka.apache.org by "Xavier Léauté (Jira)" <ji...@apache.org> on 2019/10/25 23:16:00 UTC

[jira] [Updated] (KAFKA-9106) metrics exposed via JMX shoud be configurable

     [ https://issues.apache.org/jira/browse/KAFKA-9106?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Xavier Léauté updated KAFKA-9106:
---------------------------------
    Description: 
Kafka exposes a very large number of metrics, all of which are always visible in JMX by default. On large clusters with many partitions, this may result in tens of thousands of mbeans to be registered, which can lead to timeouts with some popular monitoring agents that rely on listing JMX metrics via RMI.

Making the set of JMX-visible metrics configurable would allow operators to decide on the set of critical metrics to collect and workaround limitation of JMX in those cases.

corresponding KIP
https://cwiki.apache.org/confluence/display/KAFKA/KIP-544%3A+Make+metrics+exposed+via+JMX+configurable

  was:
Kafka exposes a very large number of metrics, all of which are always visible in JMX by default. On large clusters with many partitions, this may result in tens of thousands of mbeans to be registered, which can lead to timeouts with some popular monitoring agents that rely on listing JMX metrics via RMI.

Making the set of JMX-visible metrics configurable would allow operators to decide on the set of critical metrics to collect and workaround limitation of JMX in those cases.


> metrics exposed via JMX shoud be configurable
> ---------------------------------------------
>
>                 Key: KAFKA-9106
>                 URL: https://issues.apache.org/jira/browse/KAFKA-9106
>             Project: Kafka
>          Issue Type: Improvement
>          Components: metrics
>            Reporter: Xavier Léauté
>            Assignee: Xavier Léauté
>            Priority: Major
>
> Kafka exposes a very large number of metrics, all of which are always visible in JMX by default. On large clusters with many partitions, this may result in tens of thousands of mbeans to be registered, which can lead to timeouts with some popular monitoring agents that rely on listing JMX metrics via RMI.
> Making the set of JMX-visible metrics configurable would allow operators to decide on the set of critical metrics to collect and workaround limitation of JMX in those cases.
> corresponding KIP
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-544%3A+Make+metrics+exposed+via+JMX+configurable



--
This message was sent by Atlassian Jira
(v8.3.4#803005)