[
https://issues.apache.org/jira/browse/HADOOP-2648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12560635#action_12560635
]
stack commented on HADOOP-2648:
-------------------------------
Things look proper to me Billy. Notice how one of these regions is 'offline'
(has the additional offline:true attribute in the meta scan output). When we
split, the parent is present but offlined until the daughter regions let go of
all references (See later in the meta scan where we check if any references:
SplitA and SplitB). During this time, we'll have two regions both with same
start key but with different end keys (and different ids).
If this explains the issue you are seeing, please close this issue (Ping on IRC
if you want to learn more about how splits work).
> Region loaded on more then one region server after split
> --------------------------------------------------------
>
> Key: HADOOP-2648
> URL: https://issues.apache.org/jira/browse/HADOOP-2648
> Project: Hadoop
> Issue Type: Bug
> Components: contrib/hbase
> Reporter: Billy Pearson
> Attachments: hbase-root-master-PE1750-1.log,
> hbase-root-regionserver-PE1750-1.log, hbase-root-regionserver-PE1750-3.log
>
>
> I am uploading logs for this
> webdata has more then one region loaded that have the same starting row key
> but different end row keys
> This happens after a split sometimes they both had null as the starting row
> key
> PE1750-3 started with the table then once split the new splits went to
> PE1750-1
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
