On Tue, 17 Oct 2006 15:54:07 +0200
[log in to unmask] wrote:
> On Tue, 17 Oct 2006 14:34:06 +0100
> Matt Doidge <[log in to unmask]> wrote:
>
> > For an upgrade, would you advise a official downtime or do you think
> > things are robust enough to upgrade "on the fly".
>
> I would go for official downtime, that way if things go wrong you can
> sort them out.
>
> > And sorry to bring up the mixing 1.6/1.7 conversation again, but
> > will 1.6 pools and doors work with a 1.7 admin node? I ask because
> > if I can "spread out" the upgrading of each pool node after the
> > admin node's done then I could greatly reduce the total downtime
> > caused by the upgrade.
>
> I will see if its possible, it will take a few hours best to wait till
> tommorow. I will release the upgrade RPM in the mean time.
I wont release the upgrade script tonight, sorry will do so in morning
tommorow.
THe bad news is shut down all d-cache hosts and do the upgrade on all
hosts at the same time. We could expect streangeness and a lot of funnys
in the log if you dont. THis is now the D-cache recomended policy on
upgrade.
Regards
Owen
|