[
https://issues.apache.org/jira/browse/HDFS-6944?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jing Zhao updated HDFS-6944:
----------------------------
Attachment: HDFS-6944.001.patch
Update the patch:
# Now for a pendingMove, a target may be scheduled to the same DataNode of an
existing replica. This is because we currently use
MovedBlocks.Locations#isLocatedOn, which compares StorageGroup instances. Then
when we do the data migration the DN actually may complain that it already has
a replica and fail the migration. A fix here can be doing comparison based on
StorageGroup#getDataNodeInfo().
# Currently the Mover cannot terminate since Mover#run always returns
IN_PROGRESS. The patch adds code to wait for the existing migration finish, and
also adds a simple termination condition.
> Archival Storage: add a test framework for testing different migration
> scenarios
> --------------------------------------------------------------------------------
>
> Key: HDFS-6944
> URL: https://issues.apache.org/jira/browse/HDFS-6944
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: balancer, namenode
> Reporter: Jing Zhao
> Assignee: Jing Zhao
> Attachments: HDFS-6944.000.patch, HDFS-6944.001.patch
>
>
> This jira plans to add a testing framework for testing different scenarios of
> data migration.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
