[
https://issues.apache.org/jira/browse/HDFS-9067?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14743039#comment-14743039
]
Vinayakumar B commented on HDFS-9067:
-------------------------------------
I think, the correct fix would be to fix at the metric system level.
Say, allow metric system to support cache TTL in millis, only for tests.
May be we can have a config in "hadoop-metrics2" with name, "periodMillis" and
this will have priority over "period" which is in seconds.
So, any future tests which tests based on JMX metrics need not wait till the
metric is updated.
Default value of 'periodMillis' can be decided based on requirement, may be
100msec or even 10msec.
Any thoughts?
> o.a.h.hdfs.server.datanode.fsdataset.impl.TestLazyWriter is failing in trunk
> ----------------------------------------------------------------------------
>
> Key: HDFS-9067
> URL: https://issues.apache.org/jira/browse/HDFS-9067
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Haohui Mai
> Assignee: Surendra Singh Lilhore
> Priority: Critical
> Attachments: HDFS-9067.patch
>
>
> The test TestLazyWriter is consistently failing in trunk. For example:
> https://builds.apache.org/job/PreCommit-HDFS-Build/12407/testReport/
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
