[
https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16277175#comment-16277175
]
ramkrishna.s.vasudevan commented on HBASE-18946:
------------------------------------------------
bq.The patch is still an advance because you've moved the problem of assign to
AM which is where it belongs; that the implementation has problems is something
we can work on ...
Yes that was my concern. My early patch HBASE-18946_2 was actually working on a
completed assigned set of regions. That is why I was not looking at moving the
selection of servers to the Create table proc stage.
bq.Say some more on this please.
I was saying the same thing as you stated above advancing the target server
selection to the Create table stage and not internal to AM.
bq.Should it be other way around?
May be am missing your intent. The idea is that if tableDescriptor has
replication then create Assign procs for the entire set of regions (including
replicas) in a round robin mode with the available list of servers. If there
are no replicas just go with the existing way.
> Stochastic load balancer assigns replica regions to the same RS
> ---------------------------------------------------------------
>
> Key: HBASE-18946
> URL: https://issues.apache.org/jira/browse/HBASE-18946
> Project: HBase
> Issue Type: Bug
> Affects Versions: 2.0.0-alpha-3
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Fix For: 2.0.0-beta-1
>
> Attachments: HBASE-18946.patch, HBASE-18946.patch,
> HBASE-18946_2.patch, HBASE-18946_2.patch, HBASE-18946_simple_7.patch,
> HBASE-18946_simple_8.patch, TestRegionReplicasWithRestartScenarios.java
>
>
> Trying out region replica and its assignment I can see that some times the
> default LB Stocahstic load balancer assigns replica regions to the same RS.
> This happens when we have 3 RS checked in and we have a table with 3
> replicas. When a RS goes down then the replicas being assigned to same RS is
> acceptable but the case when we have enough RS to assign this behaviour is
> undesirable and does not solve the purpose of replicas.
> [~huaxiang] and [~enis].
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
