[
https://issues.apache.org/jira/browse/KAFKA-3480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15216310#comment-15216310
]
Jason Gustafson commented on KAFKA-3480:
----------------------------------------
[~wushujames] Feel free to pick this up if you're interested. I actually
haven't given it a lot of thought, but the JMX approach sounds a little iffy.
You'd have to also run the consumer and producer to collect their metrics and
I'm not sure all metrics are initialized upon startup (e.g. topic or node
metrics are initialized dynamically), so you'd have to have the clients do some
real work. It might also be difficult to extract patterns from the registered
metrics.
I would think doing something like ConfigDef would be a more promising approach
if you can find a way to support variable substitution in the names (e.g.
"topics.{topic}.bytes-fetched"). We'd probably want the documentation to
include the variables anyway. There is definitely some nontrivial work here,
but it would be a valuable contribution.
> Autogenerate metrics documentation
> ----------------------------------
>
> Key: KAFKA-3480
> URL: https://issues.apache.org/jira/browse/KAFKA-3480
> Project: Kafka
> Issue Type: Improvement
> Reporter: Jason Gustafson
>
> Metrics documentation is done manually, which means it's hard to keep it
> current. It would be nice to have automatic generation of this documentation
> as we have for configs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
