You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Naganarasimha G R (JIRA)" <ji...@apache.org> on 2015/09/17 09:19:46 UTC

[jira] [Commented] (YARN-4176) Resync NM nodelabels with RM every x interval for distributed nodelabels

    [ https://issues.apache.org/jira/browse/YARN-4176?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14791704#comment-14791704 ] 

Naganarasimha G R commented on YARN-4176:
-----------------------------------------

Hi [~bibinchundatt], 
Seems like this would be a better idea than what we did in YARN-4106, where in we used time interval of 1 min only on NM side failure. But i have few concerns/queries :
# I would suggest only to have only 1 resync configuration and remove what we introduced for YARN-4106.
# So node labels will be sent to RM either if node labels are modified since the last heartbeat or the resync-interval has elapsed right ?
# Earlier the way used to check the time elapse is using {{System.currentTimeMillis()}}, but i think we need to use the approach as mentioned by [~stevel@apache.org] in the [comment|https://issues.apache.org/jira/browse/HADOOP-12409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14745056#comment-14745056] of the jira HADOOP-12409. Hope [~xinxianyin] creates a new jira and provide a clock with monotonic time by then, if not use  {{System.nanoTime()}}.


> Resync NM nodelabels with RM every x interval for distributed nodelabels
> ------------------------------------------------------------------------
>
>                 Key: YARN-4176
>                 URL: https://issues.apache.org/jira/browse/YARN-4176
>             Project: Hadoop YARN
>          Issue Type: Bug
>            Reporter: Bibin A Chundatt
>            Assignee: Bibin A Chundatt
>
> This JIRA is for handling the below set of issue
> # Distributed nodelabels after NM registered with RM if cluster nodelabels are removed and added then NM doesnt resend labels in heartbeat again untils any change in labels
> # NM registration failed with Nodelabels should resend labels again to RM 
> The above cases can be handled by  resync nodeLabels with RM every x interval
> # Add property {{yarn.nodemanager.node-labels.provider.resync-interval-ms}} and  will resend nodelabels to RM based on config no matter what the registration fails or success.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)