[
https://issues.apache.org/jira/browse/HDFS-8818?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14644790#comment-14644790
]
Tsz Wo Nicholas Sze commented on HDFS-8818:
-------------------------------------------
> now that dispatcher will keep fetching more blocks from namenode every
> iteration, but namenode is likely to return very same list of blocks since
> the block moving is not that fast and namenode can't know the blocks just
> moved instantly. ...
I think it is not the case by the following reasons.
- blocksToReceive will become <= 0.
- Balancer waits until all block transfer are done each iteration and Datanodes
send a block receipt immediately once they receive a block.
- BlockManager.getBlocks(..) uses a random offset to get the blocks from the
list.
> Allow Balancer to run faster
> ----------------------------
>
> Key: HDFS-8818
> URL: https://issues.apache.org/jira/browse/HDFS-8818
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: balancer & mover
> Reporter: Tsz Wo Nicholas Sze
> Assignee: Tsz Wo Nicholas Sze
> Attachments: h8818_20150723.patch, h8818_20150727.patch
>
>
> The original design of Balancer is intentionally to make it run slowly so
> that the balancing activities won't affect the normal cluster activities and
> the running jobs.
> There are new use case that cluster admin may choose to balance the cluster
> when the cluster load is low, or in a maintain window. So that we should
> have an option to allow Balancer to run faster.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
