hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "sandflee (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-4050) NM event dispatcher may blocked by LogAggregationService if NameNode is slow
Date Thu, 12 Nov 2015 06:01:10 GMT

    [ https://issues.apache.org/jira/browse/YARN-4050?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15001734#comment-15001734
] 

sandflee commented on YARN-4050:
--------------------------------

There may be 2 problems:
1,  nm dispatcher maybe blocked by logaggregation service, should we move logaggregation event
to a new event dispatcher?
2,  nm recovery is blocked, and there're some bad effect as in YARN-4051

> NM event dispatcher may blocked by LogAggregationService if NameNode is slow
> ----------------------------------------------------------------------------
>
>                 Key: YARN-4050
>                 URL: https://issues.apache.org/jira/browse/YARN-4050
>             Project: Hadoop YARN
>          Issue Type: Bug
>            Reporter: sandflee
>
> env:  nm restart and log aggregation is enabled. 
> NN is almost dead, when we restart NM, NM event dispatcher is blocked until NN returns
to normal.It seems. NM recovered app  and send APPLICATION_START event to log aggregation
service, it will check log dir permission in HDFS(BLOCKED) 



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

Mime
View raw message