[
https://issues.apache.org/jira/browse/HBASE-20017?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16370966#comment-16370966
]
Chia-Ping Tsai commented on HBASE-20017:
----------------------------------------
{quote}I have not checked the code in detail but are you sure this is not there
in any of the newer branches after 1.4?
{quote}
Pardon me. Not only the branch-1.4 but also branch-1, branch-2, and master are
impacted. The patches I had attached are against branch-1, branch-1.4, and
master. [~apurtell] have committed the patches to all of them. There's great
shame in introducing this critical bug. :(
> BufferedMutatorImpl submit the same mutation repeatedly
> -------------------------------------------------------
>
> Key: HBASE-20017
> URL: https://issues.apache.org/jira/browse/HBASE-20017
> Project: HBase
> Issue Type: Bug
> Components: Client
> Affects Versions: 1.4.0, 1.4.1, 1.4.2
> Reporter: Chia-Ping Tsai
> Assignee: Chia-Ping Tsai
> Priority: Blocker
> Fix For: 2.0.0, 1.5.0, 1.4.2
>
> Attachments: HBASE-20017.branch-1.4.v0.patch.patch,
> HBASE-20017.branch-1.v0.patch.patch, HBASE-20017.v0.patch.patch
>
>
> BMI pass a iter of inner buffer to AsyncProcess to take the undealt
> mutations, AsyncProcess call iter#next to get the mutation and then call
> iter#remove to delete the mutation from inner buffer. Hence, There's a good
> chance that the mutation is processed repeatedly in case there are a bunch
> of threads which are running the flush.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
