[
https://issues.apache.org/jira/browse/KUDU-2007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16007372#comment-16007372
]
Mike Percy commented on KUDU-2007:
----------------------------------
Yeah that could work too. Ideally we would also make --force-copy less
necessary in the tools.
> Disallow a local-to-local tablet copy
> -------------------------------------
>
> Key: KUDU-2007
> URL: https://issues.apache.org/jira/browse/KUDU-2007
> Project: Kudu
> Issue Type: Bug
> Components: recovery
> Affects Versions: 1.3.1
> Reporter: Mike Percy
>
> Make a tablet copy from the local server to the local server always fail.
> Currently we are missing this validation check.
> Some things that might make this a little tricky:
> * On multihomed hosts it may be hard to detect that the request is local
> * In a recovery scenario, it's possible that a machine restored from backup
> could have the same UUID as an existing machine and we may want to copy a
> tablet from the tablet to the replacement, or vice versa. So they might end
> up having the same UUID.
> That said, we could probably pretty easily disallow tablet copies to / from
> the same IP/port combination and get most of our safety bang for the buck out
> of that.
> The situation where 2 hosts have the same UUID in a backup recovery situation
> might just be rare enough that we don't want to support it... but if we end
> up in such a situation then it may take custom programming to gracefully
> recover from, which would be unfortunate.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
