[
https://issues.apache.org/jira/browse/YARN-2211?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14073577#comment-14073577
]
Xuan Gong commented on YARN-2211:
---------------------------------
bq. This code not needed? may remove the newInstance() method also
It is used in RMStateStore initiation.
bq. currentKey will never be null ? if so, we can remove the check.
Might need to keep the NULL check. If the RM is start from the brand new state,
there are no states at all. So, the currentKey is NULL.
bq. Instead of moving the following to yarn_proto, we should probably have a
separate jira to move all the RM recovery related records to resource manager
module. For now, I think we can create a new proto file and move amrm token
state there.
We can do that.
> RMStateStore needs to save AMRMToken master key for recovery when RM
> restart/failover happens
> ----------------------------------------------------------------------------------------------
>
> Key: YARN-2211
> URL: https://issues.apache.org/jira/browse/YARN-2211
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager
> Reporter: Xuan Gong
> Assignee: Xuan Gong
> Attachments: YARN-2211.1.patch, YARN-2211.2.patch, YARN-2211.3.patch,
> YARN-2211.4.patch, YARN-2211.5.1.patch, YARN-2211.5.patch,
> YARN-2211.6.1.patch, YARN-2211.6.patch
>
>
> After YARN-2208, AMRMToken can be rolled over periodically. We need to save
> related Master Keys and use them to recover the AMRMToken when RM
> restart/failover happens
--
This message was sent by Atlassian JIRA
(v6.2#6252)
