[
https://issues.apache.org/jira/browse/HBASE-7191?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16748407#comment-16748407
]
stack commented on HBASE-7191:
------------------------------
[~johou] Another heuristic in hbck1 prevails, the one that considers empty dirs
as corrupt or remainders from incomplete-deletes?
You are trying to teach hbck1 to create the version and id files? The version
should be easy enough. You could create one offline and copy it into place. On
the id, I don't recall -- does it have to be just unique so any id will do or
does it have to be the id that was there previous?
> HBCK - Add offline create/fix hbase.version and hbase.id
> ---------------------------------------------------------
>
> Key: HBASE-7191
> URL: https://issues.apache.org/jira/browse/HBASE-7191
> Project: HBase
> Issue Type: Improvement
> Components: hbck
> Affects Versions: 0.94.1
> Reporter: Enis Soztutar
> Priority: Major
> Attachments: 7191-2.patch, hbck1-1.4-v1.patch
>
>
> One of our clients run into a problem, in which they have the hbase.root.dir,
> and cluster data, but their hbase.id and hbase.version files are corrupted.
> HMaster creates those on start, but not if there is already existing data.
> We can add smt like --fixIdFile, and ability for HBCK to do some offline
> repairs for the version file.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
