[
https://issues.apache.org/jira/browse/HBASE-21342?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16658283#comment-16658283
]
Ted Yu commented on HBASE-21342:
--------------------------------
bq. It is in finally block.
Perhaps I was not very clear in the previous comment.
postBulkLoadHFile() throws IOE. When this happens, I don't think Java would run
the rest of the code (decrementUgiReference in this case) in the finally block.
You can either move decrementUgiReference call to the beginning of finally
block (since it doesn't throw IOE) or, use nested finally in the current
finally block.
> FileSystem in use may get closed by other bulk load call in secure bulkLoad
> ----------------------------------------------------------------------------
>
> Key: HBASE-21342
> URL: https://issues.apache.org/jira/browse/HBASE-21342
> Project: HBase
> Issue Type: Bug
> Affects Versions: 3.0.0, 2.1.0, 1.5.0, 1.3.3, 1.4.4, 2.0.1, 1.2.7
> Reporter: mazhenlin
> Assignee: mazhenlin
> Priority: Major
> Attachments: 21342.v1.txt, HBASE-21342.002.patch,
> HBASE-21342.003.patch, HBASE-21342.004.patch, race.patch
>
>
> As mentioned in [HBASE-15291|#HBASE-15291], there is a race condition. If
> Two secure bulkload calls from the same UGI into two different regions and
> one region finishes earlier, it will close the bulk load fs, and the other
> region will fail.
>
> Another case would be more serious. The FileSystem.close() function needs two
> synchronized variables : CACHE and deleteOnExit. If one region calls
> FileSystem.closeAllForUGI ( in SecureBulkLoadManager.cleanupBulkLoad) while
> another region is trying to close srcFS ( in
> SecureBulkLoadListener.closeSrcFs) , can cause deadlock here.
>
> I have wrote a UT for this and fixed it using reference counter.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
