[
https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16280054#comment-16280054
]
ramkrishna.s.vasudevan commented on HBASE-18946:
------------------------------------------------
I can get what you are saying.
Adding to META is not an issue, we can do that any time. We can create
regioninfos from the default.
Sub procs can create new sub procs - true. But my point is this
Say first default region create a sub proc, that in turn created one sub proc
for rpelica 1. Now you call assign for that replica 1. This assign wil call
roundrobin in LB. LB will just assign to RS 1. Now again the parent sub proc
wil run - ie the default reigon. Here again round robin in LB will be called.
It will again assign RS 1 because it does not know that replica 1 is in RS 1.
I will spend some more time in this - to see if I come up with any other
alternate way. In case am wrong in the above stmts.
[~saint....@gmail.com]
You can have a go too. Sorry I missed replying to your earlier comment if you
can try this out. No problem.
> 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)
