[ https://issues.apache.org/jira/browse/YARN-4301?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15046252#comment-15046252 ] 

sandflee commented on YARN-4301:
--------------------------------

it maybe change the behaviour of NM_MIN_HEALTHY_DISKS_FRACTION, could we add a timeout to mkdir? if mkdir timeout, the disk is treated as a failed disk.

> NM disk health checker should have a timeout
> --------------------------------------------
>
>                 Key: YARN-4301
>                 URL: https://issues.apache.org/jira/browse/YARN-4301
>             Project: Hadoop YARN
>          Issue Type: Bug
>            Reporter: Akihiro Suda
>         Attachments: YARN-4301-1.patch, YARN-4301-2.patch
>
>
> The disk health checker [verifies a disk|https://github.com/apache/hadoop/blob/96677bef00b03057038157efeb3c2ad4702914da/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/DirectoryCollection.java#L371-L385] by executing {{mkdir}} and {{rmdir}} periodically.
> If these operations does not return in a moderate timeout, the disk should be marked bad, and thus {{nodeInfo.nodeHealthy}} should flip to {{false}}.
> I confirmed that current YARN does not have an implicit timeout (on JDK7, Linux 4.2, ext4) using [Earthquake|https://github.com/osrg/earthquake], our fault injector for distributed systems.
> (I'll introduce the reproduction script in a while)
> I consider we can fix this issue by making [{{NodeHealthCheckerServer.isHealthy()}}|https://github.com/apache/hadoop/blob/96677bef00b03057038157efeb3c2ad4702914da/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/NodeHealthCheckerService.java#L69-L73] return {{false}} if the value of {{this.getLastHealthReportTime()}} is too old.



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