[
https://issues.apache.org/jira/browse/HDFS-8647?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14597970#comment-14597970
]
Andrew Wang commented on HDFS-8647:
-----------------------------------
Great idea, thanks for filing this [~mingma].
Related to this, besides incorporating rack-knowledge from the BPP, we could
also consider attributes like storage type when determining if a block is
misreplicated. This would allow the NN to take over the Mover's
responsibilities and do it online rather than offline, and also edge triggered
rather than via periodic namespace scans.
This is not something we need to pursue in this JIRA, but just an idea for the
future.
> Abstract BlockManager's rack policy into BlockPlacementPolicy
> -------------------------------------------------------------
>
> Key: HDFS-8647
> URL: https://issues.apache.org/jira/browse/HDFS-8647
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Ming Ma
>
> Sometimes we want to have namenode use alternative block placement policy
> such as upgrade domains in HDFS-7541.
> BlockManager has built-in assumption about rack policy in functions such as
> useDelHint, blockHasEnoughRacks. That means when we have new block placement
> policy, we need to modify BlockManager to account for the new policy. Ideally
> BlockManager should ask BlockPlacementPolicy object instead. That will allow
> us to provide new BlockPlacementPolicy without changing BlockManager.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
