[
https://issues.apache.org/jira/browse/HDFS-1197?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13101061#comment-13101061
]
Sanjay Radia commented on HDFS-1197:
------------------------------------
The original patch dealt with ignoring the blockReceived for a BBW block sent
by a DN rebooting. However it, as a side effect, also ignores the blockReceived
sent for a block that is finalized that races ahead of the call from the client
to NN to close the block/file. Hence I think the original patch fixes a bug and
introduces another.
Jitendra observed that HDFS-1779 no longer sends the blockReceived but instead
sends a bbw-report.
Hence I think that his patch is correct. However how do you explain todd's
statement that the original patch has been in production for over a year. I
thought it would be quite common for a blockReceived for a finalized block to
race ahead of the client's call to the NN. In this case the file would not
close till the next BReport.
> Blocks are considered "complete" prematurely after commitBlockSynchronization
> or DN restart
> -------------------------------------------------------------------------------------------
>
> Key: HDFS-1197
> URL: https://issues.apache.org/jira/browse/HDFS-1197
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: data-node, hdfs client, name-node
> Affects Versions: 0.20-append
> Reporter: Todd Lipcon
> Fix For: 0.20-append
>
> Attachments: HDFS-1197-without-addStoredBlock-change.1.patch,
> hdfs-1197-test-changes.txt, hdfs-1197.txt, testTC2-failure.txt
>
>
> I saw this failure once on my internal Hudson job that runs the append tests
> 48 times a day:
> junit.framework.AssertionFailedError: expected:<114688> but was:<98304>
> at org.apache.hadoop.hdfs.AppendTestUtil.check(AppendTestUtil.java:112)
> at
> org.apache.hadoop.hdfs.TestFileAppend3.testTC2(TestFileAppend3.java:116)
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
