[
https://issues.apache.org/jira/browse/HDFS-1547?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12980930#action_12980930
]
Todd Lipcon commented on HDFS-1547:
-----------------------------------
The issue with isGoodTarget() is that it doesn't just prioritize low-load
nodes, it actively excludes any node with load > 2*avgLoad. Consider this
situation:
- 10 nodes, each of which has 2 transceivers active (total load = 20)
- decommission 6 of them
- total load is now only 8 (from the remaining ones) but "size" is still 10
- avgLoad = 0.8
- any node with load >1.6 will not be chosen
- thus all nodes are excluded and we cannot allocate a block
That is to say, decomissioning nodes now contribute to the denominator of the
total/numNodes fraction but not the numerator, so it drives the average down.
Does this make sense?
> Improve decommission mechanism
> ------------------------------
>
> Key: HDFS-1547
> URL: https://issues.apache.org/jira/browse/HDFS-1547
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: name-node
> Affects Versions: 0.23.0
> Reporter: Suresh Srinivas
> Assignee: Suresh Srinivas
> Fix For: 0.23.0
>
> Attachments: HDFS-1547.1.patch, HDFS-1547.patch
>
>
> Current decommission mechanism driven using exclude file has several issues.
> This bug proposes some changes in the mechanism for better manageability. See
> the proposal in the next comment for more details.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
