[
https://issues.apache.org/jira/browse/HDFS-13481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16443989#comment-16443989
]
Gabor Bota commented on HDFS-13481:
-----------------------------------
It seems like changing the
org.apache.hadoop.metrics2.impl.MetricsConfig#QUEUE_CAPACITY_KEY value to 2
(the default is 1) fixes the issue. The default value is 1, and sometimes
org/apache/hadoop/metrics2/impl/MetricsSinkAdapter's queue can not accept an
additional element, so RollingFileSystemSink#testFlushThread
RollingFileSystemSink is not able to flush the dropped element.
I will provide a patch soon.
> TestRollingFileSystemSinkWithHdfs#testFlushThread: test failed intermittently
> -----------------------------------------------------------------------------
>
> Key: HDFS-13481
> URL: https://issues.apache.org/jira/browse/HDFS-13481
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs
> Affects Versions: 3.0.1
> Reporter: Gabor Bota
> Assignee: Gabor Bota
> Priority: Major
>
> The test fails very rarely on a laptop, but very commonly during Jenkins runs.
> {noformat}
> Error Message
> Flush thread did not run within 10 seconds
> Stacktrace
> java.lang.AssertionError: Flush thread did not run within 10 seconds
> at
> org.apache.hadoop.metrics2.sink.TestRollingFileSystemSinkWithHdfs.testFlushThread(TestRollingFileSystemSinkWithHdfs.java:291)
> {noformat}
> According to my test, this breaks about 0.3% locally.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
