[
https://issues.apache.org/jira/browse/HDFS-16966?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17770506#comment-17770506
]
ASF GitHub Bot commented on HDFS-16966:
---------------------------------------
sodonnel commented on PR #6126:
URL: https://github.com/apache/hadoop/pull/6126#issuecomment-1741169561
Does
`dfs.datanode.available-space-volume-choosing-policy.balanced-space-preference-fraction`
not allow this to be controlled? The default is 75% of new writes go to disks
with more free space. If this is too much it could be adjust down toward 50%.
It would also be helpful if you could describe the algorithm you are
proposing as the PR has no real description about how your change working or
improves things.
> Optimize AvailableSpaceVolumeChoosingPolicy to avoid disk hotpoint
> ------------------------------------------------------------------
>
> Key: HDFS-16966
> URL: https://issues.apache.org/jira/browse/HDFS-16966
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: datanode
> Reporter: leo sun
> Priority: Major
> Labels: pull-request-available
>
> The AvailableSpaceVolumeChoosingPolicy will write the most of replicas into
> the disks with more available space than others.
> But when a new disk is launched, most write request on this DN will be on
> this disk, it will cause the disk hotpoint.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
