hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Karthik Kambatla (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-1029) Allow embedding leader election into the RM
Date Fri, 27 Dec 2013 08:41:02 GMT

    [ https://issues.apache.org/jira/browse/YARN-1029?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13857386#comment-13857386
] 

Karthik Kambatla commented on YARN-1029:
----------------------------------------

bq. Please take care of it wherever appropriate.
Re-opened YARN-1481 to take care of it there. If it isn't too much trouble, please take a
look at it. 

bq. Again, if we organize the newly added code such that its a common event for any module
to inform the RM about a fatal error then we are good for the future. Embedded elector can
use that event instead of a custom named event.
Oh! I understand it now - will add a RMFatalErrorEvent, the handler for which just terminates
the RM. And, update RMStateStoreOperationFailedEvent to use that event instead of calling
terminate directly. 

bq. I am sorry I could not understand your comment explaining how the test passes with these
timeouts.
# ZK timeout comes from RM_ZK_TIMEOUT_MS (2 seconds), the failover could take as long as this.
MiniYARNCluster#getActiveRMIndex() waits for this duration to find the active RM. 
# NM->RM connection is verified after a successful failover. The timeout there corresponds
to the maximum time taken by failovers until the NM connects to an RM. 5 seconds seems a long
enough time for this.

> Allow embedding leader election into the RM
> -------------------------------------------
>
>                 Key: YARN-1029
>                 URL: https://issues.apache.org/jira/browse/YARN-1029
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Bikas Saha
>            Assignee: Karthik Kambatla
>         Attachments: embedded-zkfc-approach.patch, yarn-1029-0.patch, yarn-1029-0.patch,
yarn-1029-1.patch, yarn-1029-2.patch, yarn-1029-3.patch, yarn-1029-4.patch, yarn-1029-approach.patch
>
>
> It should be possible to embed common ActiveStandyElector into the RM such that ZooKeeper
based leader election and notification is in-built. In conjunction with a ZK state store,
this configuration will be a simple deployment option.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message