[
https://issues.apache.org/jira/browse/HDFS-13222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16416088#comment-16416088
]
Konstantin Shvachko edited comment on HDFS-13222 at 3/27/18 6:51 PM:
---------------------------------------------------------------------
Yes you are right, the Balancer will be discarding blocks less than
{{minBlockSize}}, but it could be a lot of unnecessary blocks on a real
cluster. So the Balancer will have to send more {{getBlocks()}} requests to the
NameNode increasing unnecessarily the load on it. This will look very confusing.
I think we should fix it.
was (Author: shv):
Yes you are right, the Balancer will be discarding blocks less than
{{minBlockSize}}, but it could be a lot of unnecessary blocks on a real
cluster. So the Balancer will have to send more {{getBlocks())) requests to the
NameNode increasing unnecessarily the load on it. This will look very confusing.
I think we should fix it.
> Update getBlocks method to take minBlockSize in RPC calls
> ---------------------------------------------------------
>
> Key: HDFS-13222
> URL: https://issues.apache.org/jira/browse/HDFS-13222
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: balancer & mover
> Reporter: Bharat Viswanadham
> Assignee: Bharat Viswanadham
> Priority: Major
> Fix For: 3.1.0
>
> Attachments: HDFS-13222.00.patch, HDFS-13222.01.patch,
> HDFS-13222.02.patch
>
>
>
> getBlocks Using balancer parameter is done in this Jira HDFS-9412
>
> Pass the Balancer conf value from Balancer to NN via getBlocks in each RPC.
> as [~szetszwo] suggested
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
