[
https://issues.apache.org/jira/browse/HBASE-1224?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-1224:
-------------------------
Attachment: 1224.patch
Ben digging turned up fact that HalfMapFile sets current position back to
midkey when its the top-half of a file and its first invocation of next.
Trying to reproduce in a unit test, I see this happening but higher up in
HStoreFileScanner, while the wrong answer comes up out of the HalfMapFile, HSFS
keeps iterating the HalfMapFile until it starts returning right answers.
Something more subtle is going on. Meantime this patch makes halfmapfile do
the right thing. Maybe it fixes Bens issue.
> Scanner returns values from before startrow
> -------------------------------------------
>
> Key: HBASE-1224
> URL: https://issues.apache.org/jira/browse/HBASE-1224
> Project: Hadoop HBase
> Issue Type: Bug
> Reporter: stack
> Priority: Critical
> Fix For: 0.19.1, 0.20.0
>
> Attachments: 1224.patch
>
>
> Ben Maurer reports our returning rows from before startrow -- lots of
> concurrent scanners run out of mapreduce jobs that are updating rows as they
> go.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
