[
https://issues.apache.org/jira/browse/HDFS-13020?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Work on HDFS-13020 started by Plamen Jeliazkov.
-----------------------------------------------
> Add JMX metrics for computeReconstruction and computeInvalidation work
> ----------------------------------------------------------------------
>
> Key: HDFS-13020
> URL: https://issues.apache.org/jira/browse/HDFS-13020
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: namenode
> Reporter: Plamen Jeliazkov
> Assignee: Plamen Jeliazkov
> Priority: Minor
>
> HDFS configuration allows tweaking of NameNode replication settings. However,
> optimal settings may be tricky to determine and usually rely on observation
> of key metrics in order to determine.
> In one particular instance, I was trying to find the optimal value of
> 'dfs.namenode.replication.work.multiplier.per.iteration' but was unable to
> find any metrics around the ReconstructionMonitor's replication and
> invalidation cycles that is mostly affected by changing that property.
> If we expose some trivial JMX metrics for how long these cycles take we can
> better understand what effects tweaking the work multiplier will have.
> Of course, I am also open to suggestions around what else we should track
> here.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
