[
https://issues.apache.org/jira/browse/HDFS-3619?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13410105#comment-13410105
]
Junping Du commented on HDFS-3619:
----------------------------------
I notice one comment of isGoodBlockCandidate() in Balancer:
{noformat}
"3. doing the move does not reduce the number of racks that the block has"
{noformat}
Does this constraint looks too strong so that slightly in-consistence with
BlockPlacementPolicyDefault? In BlockPlacement, I think we only care if some
replicas are crossing the rack.
> isGoodBlockCandidate() in Balancer is not handling properly if replica factor
> >3
> --------------------------------------------------------------------------------
>
> Key: HDFS-3619
> URL: https://issues.apache.org/jira/browse/HDFS-3619
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: balancer
> Affects Versions: 1.0.0, 2.0.0-alpha
> Reporter: Junping Du
> Assignee: Junping Du
>
> Let's assume:
> 1. replica factor = 4
> 2. source node in rack 1 has 1st replica, 2nd and 3rd replica are in rack 2,
> 4th replica in rack3 and target node is in rack3.
> So, It should be good for balancer to move replica from source node to target
> node but will return "false" in isGoodBlockCandidate(). I think we can fix it
> by simply making judgement that at least one replica node (other than source)
> is on the different rack of target node.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
