[
https://issues.apache.org/jira/browse/YARN-6683?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16042294#comment-16042294
]
Rohith Sharma K S commented on YARN-6683:
-----------------------------------------
FYI to ATSv2 folks, this JIRA need not to backport to YARN-5355 branch. As part
of YARN-5638, fix is given with same approach as this jira.
> Invalid event: COLLECTOR_UPDATE at KILLED
> -----------------------------------------
>
> Key: YARN-6683
> URL: https://issues.apache.org/jira/browse/YARN-6683
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Jian He
> Assignee: Rohith Sharma K S
> Fix For: 3.0.0-alpha4
>
> Attachments: YARN-6683.001.patch, YARN-6683.001.patch
>
>
> {code}
> 2017-06-01 20:01:22,686 ERROR rmapp.RMAppImpl (RMAppImpl.java:handle(905)) -
> Can't handle this event at current state
> org.apache.hadoop.yarn.state.InvalidStateTransitionException: Invalid event:
> COLLECTOR_UPDATE at KILLED
> at
> org.apache.hadoop.yarn.state.StateMachineFactory.doTransition(StateMachineFactory.java:305)
> at
> org.apache.hadoop.yarn.state.StateMachineFactory.access$300(StateMachineFactory.java:46)
> at
> org.apache.hadoop.yarn.state.StateMachineFactory$InternalStateMachine.doTransition(StateMachineFactory.java:448)
> at
> org.apache.hadoop.yarn.server.resourcemanager.rmapp.RMAppImpl.handle(RMAppImpl.java:903)
> at
> org.apache.hadoop.yarn.server.resourcemanager.rmapp.RMAppImpl.handle(RMAppImpl.java:118)
> at
> org.apache.hadoop.yarn.server.resourcemanager.ResourceManager$ApplicationEventDispatcher.handle(ResourceManager.java:904)
> at
> org.apache.hadoop.yarn.server.resourcemanager.ResourceManager$ApplicationEventDispatcher.handle(ResourceManager.java:888)
> at
> org.apache.hadoop.yarn.event.AsyncDispatcher.dispatch(AsyncDispatcher.java:201)
> at
> org.apache.hadoop.yarn.event.AsyncDispatcher$1.run(AsyncDispatcher.java:127)
> {code}
> Below code already gets the RMApp instance and then send an event to RMApp to
> update the collector address. Instead of updating via event, it could just
> update via a method of RMApp. This also avoids state-machine changes.
> Also, is there any implications that COLLECTOR_UPDATE happened at KILLED
> state ?
> {code}
> } else {
> String previousCollectorAddr = rmApp.getCollectorAddr();
> if (previousCollectorAddr == null
> || !previousCollectorAddr.equals(collectorAddr)) {
> // sending collector update event.
> RMAppCollectorUpdateEvent event =
> new RMAppCollectorUpdateEvent(appId, collectorAddr);
> rmContext.getDispatcher().getEventHandler().handle(event);
> }
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
