[jira] [Commented] (HDFS-4768) block scanner does not close verification log when a block pool is being deleted (but the datanode remains running)

["Chris Nauroth (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HDFS-4768?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13643288#comment-13643288 ] 

Chris Nauroth commented on HDFS-4768:
-------------------------------------

This problem occasionally causes a failure in {{TestDeleteBlockPool}}.  The client request to delete the block pool races with the block scanner.  Depending on the timing, the verification log file could remain in place, and then deleting the underlying storage fails.  I can reproduce the problem more easily on Windows.

                
> block scanner does not close verification log when a block pool is being deleted (but the datanode remains running)
> -------------------------------------------------------------------------------------------------------------------
>
>                 Key: HDFS-4768
>                 URL: https://issues.apache.org/jira/browse/HDFS-4768
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: datanode
>    Affects Versions: 3.0.0, 2.0.4-alpha
>            Reporter: Chris Nauroth
>            Assignee: Chris Nauroth
>
> HDFS-4274 fixed a file handle leak of the block scanner's verification logs by adding method {{BlockPoolSliceScanner#shutdown}} and guaranteeing that the method gets called for each live {{BlockPoolSliceScanner}} during datanode shutdown.  However, that patch did not consider the case of deleting a block pool via {{ClientDatanodeProtocol#deleteBlockPool}} while the datanode remains running.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira