[
https://issues.apache.org/jira/browse/HDFS-6047?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HDFS-6047:
------------------------
Attachment: HDFS-6047.txt
Check for null chosenNode before adding it to list of nodes to ignore.
> TestPread NPE inside in DFSInputStream hedgedFetchBlockByteRange
> ----------------------------------------------------------------
>
> Key: HDFS-6047
> URL: https://issues.apache.org/jira/browse/HDFS-6047
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 2.4.0
> Reporter: stack
> Assignee: stack
> Fix For: 2.4.0
>
> Attachments: HDFS-6047.txt
>
>
> Our [~andrew.wang] saw this on internal test cluster running trunk:
> {code}
> java.lang.NullPointerException: null
> at
> org.apache.hadoop.hdfs.DFSInputStream.hedgedFetchBlockByteRange(DFSInputStream.java:1181)
> at org.apache.hadoop.hdfs.DFSInputStream.read(DFSInputStream.java:1296)
> at org.apache.hadoop.fs.FSInputStream.readFully(FSInputStream.java:78)
> at
> org.apache.hadoop.fs.FSDataInputStream.readFully(FSDataInputStream.java:108)
> at org.apache.hadoop.hdfs.TestPread.pReadFile(TestPread.java:151)
> at
> org.apache.hadoop.hdfs.TestPread.testMaxOutHedgedReadPool(TestPread.java:292)
> {code}
> TestPread was failing.
> The NPE comes of our presuming there always a chosenNode as we set up hedged
> reads inside in hedgedFetchBlockByteRange (chosenNode is null'd each time
> through the loop). Usually there is a chosenNode but need to allow for case
> where there is not.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
