[
https://issues.apache.org/jira/browse/HDFS-3689?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14091390#comment-14091390
]
Jing Zhao commented on HDFS-3689:
---------------------------------
Looks like we do not have a lot of dependency on the fixed block length. I've
found the following several places related to the preferred block size of the
file:
# concat. Currently concat has very strict preconditions including all the
corresponding files have the same preferred block size and all the files
(except the last one) should consist of full block.
# writing data. DFSOutputstream/DataStreamer allocates a new block after
writing a full block.
# append. New data is always appended to the end of the last block unless it's
just on the block boundary.
# addBlock. In the addBlock RPC, the client specifies the previous block as
null to indicate that this is the first new block for an append operation right
at the block boundary. NameNode calls {{analyzeFileState}} to verify this logic.
> Add support for variable length block
> -------------------------------------
>
> Key: HDFS-3689
> URL: https://issues.apache.org/jira/browse/HDFS-3689
> Project: Hadoop HDFS
> Issue Type: New Feature
> Components: datanode, hdfs-client, namenode
> Affects Versions: 3.0.0
> Reporter: Suresh Srinivas
> Assignee: Suresh Srinivas
>
> Currently HDFS supports fixed length blocks. Supporting variable length block
> will allow new use cases and features to be built on top of HDFS.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
