[lug] Setting up failover in Linux?

[Will]

Rob,

  Clusters are run in maintenance mode during these situations so fail
over will not occur until both systems are back to a similar state.

On Tue, Apr 24, 2012 at 3:54 PM, Rob Nagler <nagler at bivio.biz> wrote:

> In looking at various "high availability" solutions, it seems that
> software configuration is left up to the application.
>
> Consider this situation with DRBD: during a database upgrade, the
> primary fails.  What software has been installed on the secondary at
> this point?  If you installed the latest software on the primary,
> because you are doing the db upgrade with/for it, then if you are
> using DRBD on /usr, the secondary has the software.  If it doesn't,
> then it still has the old software.  Now consider this: perform a db
> upgrade, primary fails just after the commit, that is, the database is
> consistent and copied over to the secondary.  What state is the
> secondary's software?
>
> I've looked at Veritas, for example, and it seems to ignore the
> problem of software updates entirely.   Oracle doesn't deal with it
> either.
>
> What I want is a failover solution which takes into account all the
> app data and code.  I haven't seen such a solution.  Have you?
>
> Thanks,
> Rob
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lug.boulder.co.us/pipermail/lug/attachments/20120424/ec16ab7f/attachment.html>