[
https://issues.apache.org/jira/browse/HADOOP-1136?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12482916
]
Hairong Kuang commented on HADOOP-1136:
---------------------------------------
I checked the code. If a block is overreplicated, UnderReplicatedBlocks.add
handles the situation well. ArrayIndexOutOfBoundsException is impossible to be
thrown. It seems that the only possibility that causes the exception is that
the time-out block has zero copy.
> exception in UnderReplicatedBlocks:add when ther are more replicas of a block
> than required
> -------------------------------------------------------------------------------------------
>
> Key: HADOOP-1136
> URL: https://issues.apache.org/jira/browse/HADOOP-1136
> Project: Hadoop
> Issue Type: Bug
> Reporter: dhruba borthakur
> Assigned To: Hairong Kuang
>
> I was running a random writer followed by a sort when I saw this condition:
> Exception in thread "[EMAIL PROTECTED]"
> java.lang.ArrayIndexOutOfBoundsException: 3
> at
> org.apache.hadoop.dfs.FSNamesystem$UnderReplicatedBlocks.add(FSNamesystem.java:447)
> at
> org.apache.hadoop.dfs.FSNamesystem$UnderReplicatedBlocks.add(FSNamesystem.java:464)
> at
> org.apache.hadoop.dfs.FSNamesystem.processPendingReplications(FSNamesystem.java:1891)
> at
> org.apache.hadoop.dfs.FSNamesystem$ReplicationMonitor.run(FSNamesystem.java:1795)
> at java.lang.Thread.run(Thread.java:619)
> processPendingReplications is invoking neededReplications.add(). The add
> method then invokes add(block, 3, 4). This add method calls getPriority and
> that returns 3. Now, the call to priorityQueues[3].add() throws an
> OutOfBoundsException.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
