[
https://issues.apache.org/jira/browse/YARN-1028?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13823164#comment-13823164
]
Karthik Kambatla commented on YARN-1028:
----------------------------------------
When running a cluster with the patch (yarn-1028-1.patch), clients failover.
However, they take 10 seconds (try the same RM 10 times every second) before
they try the other RM. This hurts performance significantly - would be nice to
have knobs to turn that down and even better if it takes little time to
failover by default.
Does that explain the problem?
> Add FailoverProxyProvider like capability to RMProxy
> ----------------------------------------------------
>
> Key: YARN-1028
> URL: https://issues.apache.org/jira/browse/YARN-1028
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Bikas Saha
> Assignee: Karthik Kambatla
> Attachments: yarn-1028-1.patch, yarn-1028-draft-cumulative.patch
>
>
> RMProxy layer currently abstracts RM discovery and implements it by looking
> up service information from configuration. Motivated by HDFS and using
> existing classes from Common, we can add failover proxy providers that may
> provide RM discovery in extensible ways.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
