[
https://issues.apache.org/jira/browse/KAFKA-3564?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15248385#comment-15248385
]
Michael Coon commented on KAFKA-3564:
-------------------------------------
If you look at the impl for Count, it only increments by 1.0 each time I call
record (which calls update of Count from SampleStat). So it will only ever
increment my count by 1 no matter what I send in as the actual value of
record. For Total, you'll notice it overrides record and measure methods and
does NOT reset it's total, ever.
So if I want messages-in per second, I can't do new Rate(new Count()) because
it will always only increment its total by 1.0. I have sub-elements of data
coming into my processors that I want to use for the increment of count. I
would have to iterate through all of them to increment count for each
one...wasting cycles. Instead, I want to do something like
"msgsIn.record(incoming.size())" and be done with it.
If I use new Rate(new Total()), then the total value never changes so the
"rate" would just keep climbing and climbing forever and not give me the rate
of messages-in in the last minute, for example.
> Count metric always increments by 1.0
> -------------------------------------
>
> Key: KAFKA-3564
> URL: https://issues.apache.org/jira/browse/KAFKA-3564
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 0.10.1.0
> Reporter: Michael Coon
> Assignee: Kim Christensen
>
> The Count metric's update method always increments its value by 1.0 instead
> of the value passed to it. If this is by design, it's misleading as I want to
> be able to count based on values I send to the record method.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
