[
https://issues.apache.org/jira/browse/HADOOP-9700?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13706716#comment-13706716
]
Jing Zhao commented on HADOOP-9700:
-----------------------------------
bq. MAPREDUCE-2257 aims to change the copy unit to block.
Thanks for leading me to this jira, Luke! I think the incremental copy case may
be more complicated. What we need includes:
1. Changing the smallest copy unit from file to block.
2. A more powerful "concat" functionality which can concatenate two (or more)
files when the last block of the first file is not full.
3. Without 2 we can also re-transfer the last block of the old file in the
backup cluster. But in that case we may need the capability to replace a block
of a file.
bq. If you mean call createSnapshot inside discp then delete the snapshot after
job finished, I think it is reasonable.
Yes. And maybe to delete the snapshot can be optional since we may use the
snapshot for future incremental copy.
> Snapshot support for distcp
> ---------------------------
>
> Key: HADOOP-9700
> URL: https://issues.apache.org/jira/browse/HADOOP-9700
> Project: Hadoop Common
> Issue Type: New Feature
> Components: tools/distcp
> Reporter: Binglin Chang
> Assignee: Binglin Chang
> Attachments: HADOOP-9700-demo.patch
>
>
> Add snapshot incremental copy ability to distcp, so we can do iterative
> consistent backup between hadoop clusters.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
