It is possible to script this using SQL DMO/SMO and Windows WMI objects, but
I prefer to use a manual checklist. That way, there is no chance of an
accidental shutdown on the cluster. Plus, I like independent verrification
when I bring a cluster back up from a cold start that everything works, so I
add verification steps to the startup checklist. Automation is nice in
appropite circumstances, but unless you have a LOT of clusters to maintain,
this process should be a hands-on one. And by checklist, I mean precisely
that. A list with precise steps that are marked s they are completed. The
document is signed by the operator and attached to the maintenane outage
report. This list should be written and tested so the most junior staff
member with authorization can execute it successfully.
One automatic piece I found that does help is to offset the startup delay on
the Operating System so that both nodes do not come up at exactly the same
time if they power on together.
--
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP
[quoted text, click to view] "microslave" <microslave@discussions.microsoft.com> wrote in message
news:6F70DD76-0044-4EFA-ADB4-7FA2A69EC92F@microsoft.com...
> Hello Everyone,
>
> I read the useful tips about proper cluster shutdown orders. We just set
> up
> a 2 node single instance (active/passive) Win2003 / SQL 2005 cluster.
> Unfortunately our customer asks for a script or batch to shutdown the
> cluster nodes in a clean way for maintenance purposes.
> With regard to Geoff Hitens recommendations, is there any way to automate
> the shutdown procedure, including steps like setting resource startup
> types
> to DISABLED? Taking the SQL resource group offline itself or stopping the
> cluster service is easily put in a batch, but that covers only a part of
> the
> shutdown procedure.
>
> Any input will be very welcome, maybe some has come across this issue
> already?
>
> Thanks in advance,
> Roland Kiesler
Thank you for your detailed answer. I agree that a shutdown of a cluster
should involve some manual checks to control and verify this process, even if
regular shutdowns are planned, as intended by the customer. I'm currently
trying to convince them, that maintenance shutdowns of the cluster should be
handled differently compared with their shutdown procedures for their
unclustered servers and that this will require some additional (manual)
efforts. I don't feel very comfortable, thinking of scripted cluster
shutdowns. Better be safe than sorry ;-)
So thanks again, I think a tested checklist with verification steps, signed
by an operator would be a very valuable procedure to follow!
[quoted text, click to view] "Geoff N. Hiten" wrote:
> It is possible to script this using SQL DMO/SMO and Windows WMI objects, but
> I prefer to use a manual checklist. That way, there is no chance of an
> accidental shutdown on the cluster. Plus, I like independent verrification
> when I bring a cluster back up from a cold start that everything works, so I
> add verification steps to the startup checklist. Automation is nice in
> appropite circumstances, but unless you have a LOT of clusters to maintain,
> this process should be a hands-on one. And by checklist, I mean precisely
> that. A list with precise steps that are marked s they are completed. The
> document is signed by the operator and attached to the maintenane outage
> report. This list should be written and tested so the most junior staff
> member with authorization can execute it successfully.
>
> One automatic piece I found that does help is to offset the startup delay on
> the Operating System so that both nodes do not come up at exactly the same
> time if they power on together.
>
> --
> Geoff N. Hiten
> Senior Database Administrator
> Microsoft SQL Server MVP
>
>
>
> "microslave" <microslave@discussions.microsoft.com> wrote in message
> news:6F70DD76-0044-4EFA-ADB4-7FA2A69EC92F@microsoft.com...
> > Hello Everyone,
> >
> > I read the useful tips about proper cluster shutdown orders. We just set
> > up
> > a 2 node single instance (active/passive) Win2003 / SQL 2005 cluster.
> > Unfortunately our customer asks for a script or batch to shutdown the
> > cluster nodes in a clean way for maintenance purposes.
> > With regard to Geoff Hitens recommendations, is there any way to automate
> > the shutdown procedure, including steps like setting resource startup
> > types
> > to DISABLED? Taking the SQL resource group offline itself or stopping the
> > cluster service is easily put in a batch, but that covers only a part of
> > the
> > shutdown procedure.
> >
> > Any input will be very welcome, maybe some has come across this issue
> > already?
> >
> > Thanks in advance,
> > Roland Kiesler
>
Don't see what you're looking for? Try a search.
