[
https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16278462#comment-16278462
]
ramkrishna.s.vasudevan commented on HBASE-18946:
------------------------------------------------
That is what I was saying now we do the round robin decision even before. But
one thing to note is that since now the flow will go with retainAssignment it
will still go with retainAssignment what ever be the LB configured.
So stochastic LB case this will work as per the title of the JIRA.
I think even before this patch if we have some other LB configured it will
still not work as expected for replicas (and we already know that
FavouredNodeBalancer) does not take care of replicas for sure. I don't mean
this patch is solving all case and that is why I am hesitant to commit this
stil because it is not asking the LB to pick the servers. The patch _2 is
slightly better but again I think other Region balancers will not be able to
understand replica.
All replicas have been working fine only during actual balancing and not during
initial assignment.
> 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)
