[
https://issues.apache.org/jira/browse/HADOOP-1914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12529587
]
dhruba borthakur commented on HADOOP-1914:
------------------------------------------
Code look good. One comment: in the current code, the RPC layer has to
serialize a list of BlockLocations. Instead, if we create a new object that has
an array of BlockLocations inside it, and use this new object in the RPC, then
we might be able to avoid redundant bytes being serialized by the RPC layer.
> HDFS should have a NamenodeProtocol to allow secondary namenodes and
> rebalancing processes to communicate with a primary namenode
> ----------------------------------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-1914
> URL: https://issues.apache.org/jira/browse/HADOOP-1914
> Project: Hadoop
> Issue Type: New Feature
> Components: dfs
> Affects Versions: 0.14.1
> Reporter: Hairong Kuang
> Assignee: Hairong Kuang
> Fix For: 0.15.0
>
> Attachments: namenodeProtocol.patch
>
>
> For the security purpose, it is nice to have a NamenodeProtocol to allow
> secondary namenodes and rebalacing processes to communicate with a primary
> namenode. In the first version the NamenodeProtocol is going to support one
> RPC:
> BlockLocations[] getBlocks(DatanodeID datanode, long size);
> This RPC allows a rebalancing process to fetch a partial blocks map from a
> namenode. It returns a list of blocks on the given datanode and its
> locations, whose total size is the given size.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
