[jira] [Resolved] (HBASE-4350) Fix a Bloom filter bug introduced by HFile v2 and TestMultiColumnScanner that caught it

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

     [ https://issues.apache.org/jira/browse/HBASE-4350?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

stack resolved HBASE-4350.
--------------------------

      Resolution: Fixed
    Hadoop Flags: [Reviewed]

Thanks for the patch Mikhail.  Committed to TRUNK.

> Fix a Bloom filter bug introduced by HFile v2 and TestMultiColumnScanner that caught it
> ---------------------------------------------------------------------------------------
>
>                 Key: HBASE-4350
>                 URL: https://issues.apache.org/jira/browse/HBASE-4350
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Mikhail Bautin
>            Assignee: Mikhail Bautin
>             Fix For: 0.92.0
>
>         Attachments: 0001-TestMultiColumnScanner-and-Bloom-filter-fix.patch
>
>
> Nicolas pointed out to me that the new unit test TestMultiColumnScanner that I wrote for the multi-column scanner Bloom filter optimization (which we will soon release) did not pass on the open-source trunk, and it bisected down to the HFile v2 commit. I debugged the unit test and found that there was a serious bug in HFile v2 Bloom filter lookup not caught by any of the existing unit tests: Bloom filters were used for "non-Get" Scans, which did not have minimum/maximum row set correctly, and some scan results were not returned.
> This diff is the unit test that helped catch the problem and a one-line fix for the bug.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira