[
https://issues.apache.org/jira/browse/HBASE-2847?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12890086#action_12890086
]
ryan rawson commented on HBASE-2847:
------------------------------------
talking to stack it sounds like option #1 - adding in a new 8 byte long to the
KeyValue would be the most likely choice at this point. What exactly we will
store in there is still up for grabs, it should ideally be something that helps
implement ACID isolation along with this particular bug. Choices include a
region-specific per-edit counter, sequence_id or the actual timestamp.
> Put added after a delete is overshadowed if its timestamp is older than than
> that of the tombstone
> --------------------------------------------------------------------------------------------------
>
> Key: HBASE-2847
> URL: https://issues.apache.org/jira/browse/HBASE-2847
> Project: HBase
> Issue Type: Bug
> Reporter: stack
> Fix For: 0.90.0
>
>
> If we delete a row and then at a later time add to the row a cell that has a
> timestamp that is older than the delete, the addition will not be seen; the
> tombstone will prevent the newer addition being returned.
> IMO, this is non-intuitive. We should fix.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
