[
https://issues.apache.org/jira/browse/KAFKA-12563?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17309191#comment-17309191
]
Ryanne Dolan commented on KAFKA-12563:
--------------------------------------
The metric is calculated based on the timestamp of each replicated record,
which can be set by client code. e.g. a producer can set a timestamp of zero,
which would yield the replication latency you observe.
> Something wrong with MM2 metrics
> --------------------------------
>
> Key: KAFKA-12563
> URL: https://issues.apache.org/jira/browse/KAFKA-12563
> Project: Kafka
> Issue Type: Bug
> Components: mirrormaker
> Affects Versions: 2.7.0
> Reporter: Bui Thanh MInh
> Priority: Major
> Attachments: Screen Shot 2021-03-26 at 12.10.12.png
>
>
> The metric
> _*`adt_2dc_c1_kafka_connect_mirror_source_connector_replication_latency_ms_avg`*_
> shows that value of latency is a very large number but the amount of
> messages in two DC are the same.
> View details in the attachment.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
