[
https://issues.apache.org/jira/browse/HDFS-9330?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15118352#comment-15118352
]
Arpit Agarwal commented on HDFS-9330:
-------------------------------------
Thanks for the updated patch [~xiaobingo]. My comments:
# The changes can be simplified by moving the {{deleteDuplicateReplicas}} flag
into FsDatasetImpl. It can be passed to {{resolveDuplicateReplicas}} and
{{addReplicaToReplicasMap}} as a parameter. Then you don't need the logic to
enumerate block pool slices in FsDatasetImpl. Also avoids changes to
FsVolumeImpl.
# {{deleteDuplicateReplicas}} can be a {{volatile boolean}} instead of
{{AtomicBoolean}} as we don't need compare-and-swap semantics.
# {{FsDatasetImpl#getNewConf}} should throw an exception.There is no reason for
anyone to call it.
# Replace calls to {{FsDatasetImpl#getConf}} with {{DataNode#getConf}} as only
latter can return the active configuation.
# Datanode.java:566 - redundant cast to Reconfigurable.
Couple of pre-existing bugs, not introduced by your patch:
# {{DataNode#refreshNamenodes}} should instantiate {{HdfsConfiguration}}. We
can fix it here as it is a trivial change.
# Configuration objects should no longer be cached in the DataNode e.g. in
DNConf.java. We can fix it in a separate Jira.
> Support reconfiguring dfs.datanode.duplicate.replica.deletion without DN
> restart
> ---------------------------------------------------------------------------------
>
> Key: HDFS-9330
> URL: https://issues.apache.org/jira/browse/HDFS-9330
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: datanode
> Reporter: Xiaobing Zhou
> Assignee: Xiaobing Zhou
> Attachments: HDFS-9330-HDFS-9000.002.patch,
> HDFS-9330-HDFS-9000.003.patch, HDFS-9330.001.patch
>
>
> This is to reconfigure
> {code}
> dfs.datanode.duplicate.replica.deletion
> {code}
> without restarting DN.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
