[
https://issues.apache.org/jira/browse/HBASE-18703?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16175318#comment-16175318
]
Umesh Agashe edited comment on HBASE-18703 at 9/21/17 7:23 PM:
---------------------------------------------------------------
[~anoop.hbase], [~Apache9]:
These two code paths look very similar and I found comments/ todos in the code
about unifying these two paths. The main difference I see here is:
* one is atomic and other is not
* Atomic path guarantees the oder of operations and non-atomic path does not
To remove inconsistencies (above mentioned + a few other minor), I am working
on a patch to unify these paths. I think I will be able to upload the patch
soon. Let me know your thoughts.
was (Author: uagashe):
[~anoopamz], [~Apache9]:
These two code paths look very similar and I found comments/ todos in the code
about unifying these two paths. The main difference I see here is:
* one is atomic and other is not
* Atomic path guarantees the oder of operations and non-atomic path does not
To remove inconsistencies (above mentioned + a few other minor), I am working
on a patch to unify these paths. I think I will be able to upload the patch
soon. Let me know your thoughts.
> Inconsistent behavior for preBatchMutate in doMiniBatchMutate and
> processRowsWithLocks
> --------------------------------------------------------------------------------------
>
> Key: HBASE-18703
> URL: https://issues.apache.org/jira/browse/HBASE-18703
> Project: HBase
> Issue Type: Sub-task
> Components: Coprocessors
> Reporter: Duo Zhang
> Assignee: Umesh Agashe
> Priority: Critical
> Fix For: 2.0.0-alpha-4
>
>
> In doMiniBatchMutate, the preBatchMutate is called before building WAL, but
> in processRowsWithLocks, we suggest the RowProcessor implementation to build
> WAL in process method, which is ahead of preBatchMutate.
> If a CP modifies the mutations, especially if it removes some cells from the
> mutations, then the behavior of processRowsWithLocks is broken. The changes
> applied to memstore and WAL will be different. And there is no way to remove
> entries from a WALEdit through CP.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
