[
https://issues.apache.org/jira/browse/KAFKA-10217?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17152034#comment-17152034
]
Bruno Cadonna commented on KAFKA-10217:
---------------------------------------
To use {{nodeLevelSensor()}} in {{StreamsMetrics}} we would also need to expose
the node name publicly. Currently, we only expose it in
{{InternalProcessorContext}} via {{currentNode()}}.
> Move nodeLevelSensor and storeLevelSensor methods from StreamsMetricsImpl to
> StreamsMetrics
> -------------------------------------------------------------------------------------------
>
> Key: KAFKA-10217
> URL: https://issues.apache.org/jira/browse/KAFKA-10217
> Project: Kafka
> Issue Type: Improvement
> Components: streams
> Reporter: John Roesler
> Assignee: mohamed chebbi
> Priority: Minor
> Labels: kip-required, newbie
>
> StreamsMetricsImpl contains several convenience methods for safely
> registering sensors at different levels of granularity. We added them as
> internal methods because we weren't sure of their stability and utility. Now,
> they've been in use for quite a while and seem to be stable.
> We should move them up into the public API so that custom stores and
> processor implementations can also benefit from their safety.
> Implementing this feature should also allow us to drop the adaptor function:
> `org.apache.kafka.streams.processor.internals.ProcessorContextUtils#getMetricsImpl`
> Note: this feature requires a KIP, but since the API is already
> pre-determined, it should be uncontroversial. This improvement would be a
> good opportunity for someone who wants to get an initial KIP under their belt.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
