[
https://issues.apache.org/jira/browse/HBASE-12751?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14377157#comment-14377157
]
Elliott Clark commented on HBASE-12751:
---------------------------------------
So [~nkedel] and I have been debugging this today. The test that's failing is a
real test failure. However it's do to the late binding sequence id work that
went in a while ago.
The issue is that if we allow two WALEdits for the same row to be outstanding
then it's possible that the pre-created sequence number (1M + a incrementing
long) will not sort in the same way as the sequence id inside of the waledit.
This late binding is currently pretty intwined with the FSHLog.
[~stack] or [~busbey] or [~jeffreyz] any thoughts on how to fix this cleanly?
My only thought was to move memstore insertion after hlog.append.
> Allow RowLock to be reader writer
> ---------------------------------
>
> Key: HBASE-12751
> URL: https://issues.apache.org/jira/browse/HBASE-12751
> Project: HBase
> Issue Type: Bug
> Components: regionserver
> Reporter: Elliott Clark
> Assignee: Nate Edel
> Attachments: HBASE-12751-v1.patch, HBASE-12751.patch
>
>
> Right now every write operation grabs a row lock. This is to prevent values
> from changing during a read modify write operation (increment or check and
> put). However it limits parallelism in several different scenarios.
> If there are several puts to the same row but different columns or stores
> then this is very limiting.
> If there are puts to the same column then mvcc number should ensure a
> consistent ordering. So locking is not needed.
> However locking for check and put or increment is still needed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
