hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ramkrishna.s.vasudevan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-18946) Stochastic load balancer assigns replica regions to the same RS
Date Mon, 04 Dec 2017 09:11:00 GMT

    [ https://issues.apache.org/jira/browse/HBASE-18946?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16276483#comment-16276483
] 

ramkrishna.s.vasudevan commented on HBASE-18946:
------------------------------------------------

bq.But the problem is no longer we call roundrobinAssignment from LB instead we will call
retainAssignment() only. This happens only when replicas are created
I said this because now we set the target server so the round robin logic is not with the
AM. Generally AM decides if the assignment shouldbe round robin or retain assignment. So here
it is always retain assignment. 
As I said LB knows about replica and it does the replica assignment pretty well but that is
when it has the entire cluster available. During the course of assignment it cannot make a
judgement on it. I don't find any other simple way to make LB aware of it unless some where
we need a grouping of the regions to be assigned. So are you ok with patch for commit [~saint.ack@gmail.com]?


> 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)

Mime
View raw message