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 "Xuan Gong (JIRA)" <ji...@apache.org> on 2014/01/31 02:13:08 UTC

[jira] [Commented] (YARN-1611) Make admin refresh of scheduler configuration work across RM failover

    [ https://issues.apache.org/jira/browse/YARN-1611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13887339#comment-13887339 ] 

Xuan Gong commented on YARN-1611:
---------------------------------

bq. Fix formatting in the code. There are serveral non-standard instances of formatting, and lines crossing 80 char boundary.

DONE

bq. conf.store -> remote-configuration.store

changed

bq. Change DEFAULT_RM_CONF_STORE to be something like /yarn/conf?

changed 

bq.Add javadoc for RemoteConfiguration the class and all the methods.

added

bq. Same for RemoteConfigurationFactory.

added

bq. Move RC and RCF to yarn.conf package

Moved

bq. FileSystemBasedRemoteConfiguration: If path doesn't exist, we should not silently log it. Throw an exception.

changed

bq.CapacityScheduler.java: Remote conf is loaded only on refresh but not on init?

Yes, we need to do this. But this will be fixed in YARN-1459

bq. Instead of getConfigurationFileName(), create static constants for each config-file and directly code them into the caller.

DONE. Just create CS_CONFIGURATION_FILE in this ticket. Will create other conf file within related jira tickets

bq. When can conf be null in HA mode? Even if it can, the response to refresh to indicate an exception.

If we throw out exception in FileSystemBasedRemoteConfiguration If path doesn't exist. So, the conf will not be null. Changed.

> Make admin refresh of scheduler configuration work across RM failover
> ---------------------------------------------------------------------
>
>                 Key: YARN-1611
>                 URL: https://issues.apache.org/jira/browse/YARN-1611
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Xuan Gong
>            Assignee: Xuan Gong
>         Attachments: YARN-1611.1.patch, YARN-1611.2.patch, YARN-1611.2.patch, YARN-1611.3.patch, YARN-1611.3.patch, YARN-1611.4.patch, YARN-1611.5.patch, YARN-1611.6.patch
>
>
> Currently, If we do refresh* for a standby RM, it will failover to the current active RM, and do the refresh* based on the local configuration file of the active RM. 



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