[
https://issues.apache.org/jira/browse/HDFS-957?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12831803#action_12831803
]
Konstantin Shvachko commented on HDFS-957:
------------------------------------------
# I don't understand what problem this solves. We always save new image into
IMAGE_NEW. Then we rename it to IMAGE. If IMAGE_NEW exists we treat it as if
the save did not complete (unsuccessful checkpoint or save). IMAGE_NEW then is
simply removed and the old image in IMAGE file is used as the current image.
There is no need to track progress of writing into IMAGE_NEW.
# It is very important in upgrade that we do not write anything into VERSION
file until IMAGE file is written. Writing VERSION file is like a commit when
the image directory is formed. We first write everything else then VERSION
file, which indicates the directory is in good order.
If you create VERSION file before completing the image all this logic will not
work.
> FSImage layout version should be only once file is complete
> -----------------------------------------------------------
>
> Key: HDFS-957
> URL: https://issues.apache.org/jira/browse/HDFS-957
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: name-node
> Affects Versions: 0.22.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Attachments: hdfs-957.txt
>
>
> Right now, the FSImage save code writes the LAYOUT_VERSION at the head of the
> file, along with some other headers, and then dumps the directory into the
> file. Instead, it should write a special IMAGE_IN_PROGRESS entry for the
> layout version, dump all of the data, then seek back to the head of the file
> to write the proper LAYOUT_VERSION. This would make it very easy to detect
> the case where the FSImage save got interrupted.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
