You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by "Jim Kellerman (JIRA)" <ji...@apache.org> on 2008/07/18 19:44:31 UTC
[jira] Commented: (HBASE-753) Can't replace the data of a
particular table by copying its files on HDFS
[ https://issues.apache.org/jira/browse/HBASE-753?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12614818#action_12614818 ]
Jim Kellerman commented on HBASE-753:
-------------------------------------
You can't just copy a table's directories and expect it to work. It might if all the regions are the same, but if a region has split, data in the target META won't have this information.
Maybe when we get hbase-fsck you'll be able to do it if you run the fsck after the copy and before starting the cluster. Until that time, you'll either get lucky and it will work, or it won't.
> Can't replace the data of a particular table by copying its files on HDFS
> -------------------------------------------------------------------------
>
> Key: HBASE-753
> URL: https://issues.apache.org/jira/browse/HBASE-753
> Project: Hadoop HBase
> Issue Type: Bug
> Affects Versions: 0.2.0
> Reporter: Sebastien Rainville
> Priority: Minor
>
> I have 2 instances of hbase running. One is *production* and the other one is *development*. I want to be able to replace the content of a table (not all of them) in development by the content in production. Both of my environments are running hbase-trunk (a snapshot of july 9th). In hbase-0.1.x we used to be able to do that by simply stopping both hbases, copying the files of the required table directly from one HDFS to the other and then restart hbase.
> It doesn't work anymore. In hbase shell I do see the table but it's empty.
> There are no errors. I looked at the master's log and the regionservers logs as well, all in DEBUG mode... but I saw nothing interesting. I do see that the regions for that table are being assigned. So, if there's more than 1 region it means that it knows that the table isn't empty.
> So, I have to copy all the tables and then it's fine. It's not practical though.
> My guess is that .META. is holding old information about that table that doesn't get updated when I replace the table's data.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.