[
https://issues.apache.org/jira/browse/HADOOP-6918?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13029600#comment-13029600
]
Luke Lu commented on HADOOP-6918:
---------------------------------
bq. This jira seems like the best one to me since it's also flagged
incompatible.
Make sense, reopen.
As for workaround, metrics sinks can be trivially enhanced to have user defined
mappings, which is what we did for production deployment.
> Make metrics naming consistent
> ------------------------------
>
> Key: HADOOP-6918
> URL: https://issues.apache.org/jira/browse/HADOOP-6918
> Project: Hadoop Common
> Issue Type: Improvement
> Components: metrics
> Affects Versions: 0.20.2
> Reporter: Luke Lu
> Assignee: Luke Lu
> Fix For: 0.23.0
>
>
> While working HADOOP-6728, I noticed that our metrics naming style is all
> over the place:
> * Capitalized camel case: e.g., "FilesCreated" in namenode metrics and some
> rpc metrics
> * uncapitalized camel case: e.g, "threadsBlocked" in jvm metrics and some rpc
> metrics
> * lowercased underscored: e.g., "bytes_written" in datanode metrics and
> mapreduce metrics
> Let's make them consistent. How about uncapitalized camel case? My main
> reason for the camel case: some backends have limits on the name length and
> underscore is wasteful.
> Once we have a consistent naming style we can do:
> @Metric("Number of INodes created") MutableCounterLong filesCreated;
> instead of the more redundant:
> @Metric({"FilesCreated", "Number of INodes created"}) MutableCounterLong
> filesCreated;
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
