You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Jess (Jira)" <ji...@apache.org> on 2025/07/29 18:09:00 UTC
[jira] [Updated] (KAFKA-19532) Kafka request metric discrepancies
[ https://issues.apache.org/jira/browse/KAFKA-19532?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jess updated KAFKA-19532:
-------------------------
Description:
Following up on the discussion in this Kafka discussion thread: [https://lists.apache.org/thread/w4n0tbpn08ocnyq51vj0mjhxmbn42fcv|https://lists.apache.org/thread/w4n0tbpn08ocnyq51vj0mjhxmbn42fcv.]
It appears that there is a possible bug between the metrics `kafka.network:type=RequestMetrics,name=RequestsPerSec,request= \{Produce|FetchConsumer|FetchFollower},version=([0-9]+)` and these two metrics:
* `kafka.server:type=BrokerTopicMetrics,name=TotalProduceRequestsPerSec,topic=([-.\\w]+)`
* `kafka.server:type=BrokerTopicMetrics,name=TotalFetchRequestsPerSec,topic=([-.\\w]+)`
Both represent requests per second, which makes it especially confusing when I see completely different values for both. I would assume data would line up for these metrics, so maybe I am misunderstanding how these metrics work (`kafka.network` vs `kafka.server`) or whether there is simply a bug (as suggested by the discussion thread) but I would like to get to the bottom of this confusion.
was:
Following up on the discussion in this Kafka discussion thread: [https://lists.apache.org/thread/w4n0tbpn08ocnyq51vj0mjhxmbn42fcv.]
It appears that there is a possible bug between the metrics `kafka.network:type=RequestMetrics,name=RequestsPerSec,request= \{Produce|FetchConsumer|FetchFollower},version=([0-9]+)` and these two metrics:
* `kafka.server:type=BrokerTopicMetrics,name=TotalProduceRequestsPerSec,topic=([-.\\w]+)`
* `kafka.server:type=BrokerTopicMetrics,name=TotalFetchRequestsPerSec,topic=([-.\\w]+)`
Both represent requests per second, which makes it especially confusing when I see completely different values for both. I would assume data would line up for these metrics, so maybe I am misunderstanding how these metrics work (`kafka.network` vs `kafka.server`) or whether there is simply a bug (as suggested by the discussion thread) but I would like to get to the bottom of this confusion.
> Kafka request metric discrepancies
> ----------------------------------
>
> Key: KAFKA-19532
> URL: https://issues.apache.org/jira/browse/KAFKA-19532
> Project: Kafka
> Issue Type: Bug
> Reporter: Jess
> Assignee: Ksolves India Limited
> Priority: Minor
>
> Following up on the discussion in this Kafka discussion thread: [https://lists.apache.org/thread/w4n0tbpn08ocnyq51vj0mjhxmbn42fcv|https://lists.apache.org/thread/w4n0tbpn08ocnyq51vj0mjhxmbn42fcv.]
> It appears that there is a possible bug between the metrics `kafka.network:type=RequestMetrics,name=RequestsPerSec,request= \{Produce|FetchConsumer|FetchFollower},version=([0-9]+)` and these two metrics:
> * `kafka.server:type=BrokerTopicMetrics,name=TotalProduceRequestsPerSec,topic=([-.\\w]+)`
> * `kafka.server:type=BrokerTopicMetrics,name=TotalFetchRequestsPerSec,topic=([-.\\w]+)`
> Both represent requests per second, which makes it especially confusing when I see completely different values for both. I would assume data would line up for these metrics, so maybe I am misunderstanding how these metrics work (`kafka.network` vs `kafka.server`) or whether there is simply a bug (as suggested by the discussion thread) but I would like to get to the bottom of this confusion.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)