[
https://issues.apache.org/jira/browse/KAFKA-541?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yang Ye updated KAFKA-541:
--------------------------
Attachment: kafka_541_v3.diff
metrics-annotation-3.0.1.jar
metrics-core-3.0.1.jar
changes since v2:
1. patch the coda hale metrics project and build new jars, which fixes the
csvReporter not reporting meter attributes for timer metrics problem
2. fix the metric.py by not passing non-existent files into inputCsvFiles of
plot_graphs() function
3. change the metrics in kafka RequestMetrics class from in unit of nano
seconds to milli seconds
4. build a getter for "props" attribute in ConsumerConfig,
5. making the KafkaCSVMetricsReporter a singleton, only one instance can be
started within each JVM
> Use metrics CSV reporter instead of jmx tool for system tests
> -------------------------------------------------------------
>
> Key: KAFKA-541
> URL: https://issues.apache.org/jira/browse/KAFKA-541
> Project: Kafka
> Issue Type: Bug
> Affects Versions: 0.8
> Reporter: Joel Koshy
> Assignee: Yang Ye
> Attachments: kafka_541_v1.diff, kafka_541_v2.diff, kafka_541_v3.diff,
> metrics-annotation-3.0.1.jar, metrics-core-3.0.1.jar
>
>
> The existing system test framework spawns off a bunch of jmxtool processes to
> collect metrics. This is rather heavy-weight and also, requires advance
> knowledge of all the beans (many of which are dynamically registered). E.g.,
> per-topic stats pop-up only after the topics are produced to.
> Since we are using metrics-core, we can just turn on the CSV reporter to
> collect these stats. I had originally thought version 2.1.3 had various bugs
> that rendered it unusable for CSV reporter, but I gave it another try and it
> seems to be fine. Will post some output.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
