[
https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16264225#comment-16264225
]
ramkrishna.s.vasudevan commented on HBASE-18946:
------------------------------------------------
bq.Man. We want round-robin but with caveats.. "Round-robin except..."
I think you are worried about this part. But I see this way.
We are still doing round robin.
1) Assign primary replicas (round robin)
2) assign sec replicas ( again round robin just avoid primary replica servers).
3) assign tertiary replicas (again round robin but avoid servers in first two
steps).
and so on. ..
So you think why not do a round robin with the entire set of regions. That is
true round robin. But for that some where we need to hold up the regions and
bulk it up before going for Assign procs.
Otherwise another thing we can do is that,
Add a API in AM that will do tentative round robin with the available servers
and regions. And in the CreateTableProc before creating the AssignProcs add the
target server also to it. So that we go ahead and assign accordingly only.
bq.Could we pass the AM the new table regions and ask it to return us plans to
use assigning?
Reading this comment of yours -you suggest the same I believe? Can we do that
only for tables with replicas?
> 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,
> 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)
