[
https://issues.apache.org/jira/browse/YARN-1028?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13823126#comment-13823126
]
Karthik Kambatla commented on YARN-1028:
----------------------------------------
I propose the following:
# Introduce {[ipc.client.connect.retry.interval}} and set it to 1 second by
default
# Introduce {{yarn.resourcemanager.ha.failover-interval-ms}} with a default
value of 1 second to capture how soon to failover. (Should we reuse
{{yarn.resourcemanager.connect.retry-interval.ms}} for this also?) Based on
this and the value of {{ipc.client.connect.max.retries}}, dynamically set
{{ipc.client.connect.retry.interval}}. For example, for the default values of 1
second for failover-interval and 10 for max.retries, the
{{ipc.client.connect.retry.interval}} becomes 100 ms.
# The number of times to failover is to be bounded by
{{yarn.resourcemanager.connect.max-wait.ms}}
[~bikassaha], what do you think?
> 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)
