[
https://issues.apache.org/jira/browse/TEZ-2311?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14647209#comment-14647209
]
Hitesh Shah commented on TEZ-2311:
----------------------------------
Mostly looks good.
One comment though:
{code}
if (recoveredDAGData.isSessionStopped) {
1831 LOG.info("AM crashed when shutting down in the previous attempt"
1832 + ", continue the shutdown and recover it to SUCCEEDED");
1833 this.state = DAGAppMasterState.SUCCEEDED;
1834 this.taskSchedulerEventHandler.setShouldUnregisterFlag();
1835 shutdownHandler.shutdown();
1836 return;
1837 }
{code}
Should the above only be done for session mode? For non-session mode, the final
state should be set to KILLED. Also, I don't see "sessionStopped" set to false
anywhere.
One option for non-session mode is to let the dag recover as needed and then
the AM will shutdown once the dag finished event comes back? Or just invoke
shutdown before that happens.
> AM can hang if kill received while recovering from previous attempt
> -------------------------------------------------------------------
>
> Key: TEZ-2311
> URL: https://issues.apache.org/jira/browse/TEZ-2311
> Project: Apache Tez
> Issue Type: Bug
> Affects Versions: 0.6.0
> Reporter: Jason Lowe
> Assignee: Jeff Zhang
> Labels: Recovery
> Attachments: TEZ-2311-1.patch, TEZ-2311-2.patch, TEZ-2311-3.patch,
> TEZ-2311-4.patch, TEZ-2311-5.patch, TEZ-2311-6.patch
>
>
> We saw an instance of a Tez job hanging despite receiving multiple kill
> requests from clients. The AM was recovering from a prior attempt when the
> first kill request arrived.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
