[
https://issues.apache.org/jira/browse/MAPREDUCE-5631?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13826960#comment-13826960
]
Jonathan Eagles commented on MAPREDUCE-5631:
--------------------------------------------
Good thought, Jason. After further investigation, the failing test is using a
Configuration object. If it runs first, it passes since none of the
{mapred,yarn}-{default,site}.xml and assumes the source code default retry
interval of 5 seconds. If it runs after any of the test that run a job, it
fails since the first use of JobConf will add the default and site files to the
default resources of the configuration class. The assumption of falling back to
the source code default of 5 seconds is no longer valid, since it is picking of
the mapred-default.xml value of 1 second.
I will change this test class to use the jobconf object to avoid any test order
dependencies and add the config settings explicit to avoid changes to the
default.xml causing the test to fail again.
> TestJobEndNotifier.testNotifyRetries fails with Should have taken more than 5
> seconds in jdk7
> ---------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-5631
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5631
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Affects Versions: 3.0.0, 2.3.0, 0.23.10
> Reporter: Jonathan Eagles
> Assignee: Jonathan Eagles
> Labels: java7
> Attachments: MAPREDUCE-5631.patch
>
>
> Configuration settings are bleeding over from test to test in jdk7
> environment since tests are run in random order.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
