[jira] Updated: (HBASE-564) Adding a flush file of zero entries

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

     [ https://issues.apache.org/jira/browse/HBASE-564?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jim Kellerman updated HBASE-564:
--------------------------------

    Attachment: patch.txt

There were thee paths that bypassed the size checks:
- HStore.doReconstructionLog called by HStore constructor
- HStore.flushCache called by HRegion.internalFlushCache called by either HRegion.close or HRegion.flushCache

Added a check in HStore.internalFlushCache to just return if cache has zero entries.


> Adding a flush file of zero entries
> -----------------------------------
>
>                 Key: HBASE-564
>                 URL: https://issues.apache.org/jira/browse/HBASE-564
>             Project: Hadoop HBase
>          Issue Type: Bug
>            Reporter: stack
>            Assignee: Jim Kellerman
>            Priority: Minor
>         Attachments: patch.txt
>
>
> Saw this in log in TRUNK:
> {code}
>     [junit] 2008-04-04 20:22:40,943 DEBUG [RegionServer:0.cacheFlusher] regionserver.HStore(676): Added 1403560700/text/8075392345773720818 with 0 entries, sequence id 537, data size 0.0, file size 110.0 for 1403560700/text
> {code}
> I thought that we'd fixed flushing zero-entry files

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.