[
https://issues.apache.org/jira/browse/HDFS-231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13066716#comment-13066716
]
Harsh J commented on HDFS-231:
------------------------------
I'd say that the HDFS is such a system that you do not really require backups
as long as you have active node monitoring; and for a few files its alright if
you do it non-incrementally.
For instance, you can also look at dates.
> Rsync like way of retrieving data from the dfs
> ----------------------------------------------
>
> Key: HDFS-231
> URL: https://issues.apache.org/jira/browse/HDFS-231
> Project: Hadoop HDFS
> Issue Type: New Feature
> Reporter: Johan Oskarsson
> Assignee: Sameer Paranjpye
>
> As the dfs in our cluster contains a lot of important data, being able to
> retrieve them to a non dfs backup node is essential.
> However, a lot of the files don't change inbetween backups, so a way to get
> only the files that have changed would be preferable.
> Since the blocks themselves already have a crc calculated half the job is
> already done, if it's possible to split the destination files in similar
> blocks and calculate the crc for them.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
