[
https://issues.apache.org/jira/browse/HBASE-1014?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12650419#action_12650419
]
stack commented on HBASE-1014:
------------------------------
Slava: I'm trying to think of a case where it would be critical that the
timestamping is done on the server-side rather than in client? If you have
one, stick it in here. My thinking is that even if I came up with one, what
about case where server crashes and region gets deployed to another server and
its clock is off with the original regionservers? Then even though the
timestamping is being done serverside, it would be a mistake to depend on it
always being right? And IIRC, regards the transactional hbase, the client runs
the transaction so again it should be ok if its setting timestamps?
> commit(BatchUpdate) method should return timestamp
> ---------------------------------------------------
>
> Key: HBASE-1014
> URL: https://issues.apache.org/jira/browse/HBASE-1014
> Project: Hadoop HBase
> Issue Type: Improvement
> Reporter: Slava
> Fix For: 0.19.0
>
>
> The commit(BatchUpdate) and commit(list<BatchUpdate>) should return timestamp
> that BatchUpdate was committed with (in the case of commit(list<BatchUpdate>
> should return array of timestamps).
> This should reduce number of round trips and improve performance in update
> operations.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
