[ 
https://issues.apache.org/jira/browse/HBASE-20049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16374151#comment-16374151
 ] 

Hudson commented on HBASE-20049:
--------------------------------

FAILURE: Integrated in Jenkins build HBase-Trunk_matrix #4634 (See 
[https://builds.apache.org/job/HBase-Trunk_matrix/4634/])
HBASE-20049 Region replicas of SPLIT and MERGED regions are kept in (tedyu: rev 
1bc996aa505e4a064a5bdf6abd56f77591643072)
* (edit) 
hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/UnassignProcedure.java
* (edit) 
hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/MergeTableRegionsProcedure.java
* (edit) 
hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/SplitTableRegionProcedure.java
* (edit) hbase-protocol-shaded/src/main/protobuf/MasterProcedure.proto
* (edit) 
hbase-server/src/main/java/org/apache/hadoop/hbase/master/assignment/AssignmentManager.java


> Region replicas of SPLIT and MERGED regions are kept in in-memory states 
> until restarting master
> ------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-20049
>                 URL: https://issues.apache.org/jira/browse/HBASE-20049
>             Project: HBase
>          Issue Type: Bug
>          Components: read replicas
>            Reporter: Toshihiro Suzuki
>            Assignee: Toshihiro Suzuki
>            Priority: Major
>             Fix For: 2.0.0-beta-2
>
>         Attachments: HBASE-20049.master.001.patch, 
> HBASE-20049.master.002.patch, HBASE-20049.master.003.patch
>
>
> Currently, when regions are split or merged, region replicas of the regions 
> will become CLOSED state but will not be deleted from in-memory states (e.g. 
> AssignmentManager). These region replicas are deleted when restarting master, 
> but I think they should be deleted during splitting and merging regions. 
> One problem with this is the outdated region replicas can be online. Steps to 
> reproduce are as follows:
> 1. Create a table with enabling region replication.
> {code}
> create "test", "cf", {REGION_REPLICATION => 2}
> {code}
> 2. Load data to the table
> {code}
> (0...2000).each{|i| put "test", "row#{i}", "cf:col", "val"}
> {code}
> 3. Split the table
> {code}
> split "test"
> {code}
> 4. Disable and enable the table
> {code}
> disable "test"
> enable "test"
> {code}
> After doing those steps, the outdated region replica will be online.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to