Re: Properly shutdown Windows 2003 Cluster during emergency power

Tech-Archive recommends: Fix windows errors by optimizing your registry

Thanks Rod!

"Rodney R. Fournier [MVP]" wrote:

> Losing power completely across the cluster could be very bad. Data could be
> lost, depends on your RAID controllers and SAN architecture. You should not
> lose anything, but it can and does happen. I have even seen the Quorum get
> corrupted because of an improper shutdown. A UPS should be mandatory for
> every cluster.
>
> With the boot.ini, you really want a node to come up and control the
> cluster. This will make the rest of the cluster come up smoothly. It does
> not matter who owned the cluster before, which ever nodes come up first,
> will own it upon reboot.
>
> Cheers,
>
> Rod
>
> MVP - Windows Server - Clustering
> http://www.nw-america.com - Clustering Website
> http://www.msmvps.com/clustering - Blog
>
> "Doug" <Doug@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> news:CBA8A66F-73D5-45FF-8482-AE39F54006EF@xxxxxxxxxxxxxxxx
> > Rodney,
> >
> > Thanks for all the great feedback! Yes we do have the software configured
> > to
> > shutdown the servers within 5 min of power loss. We also stagger the power
> > up
> > in boot.ini just like you had suggested. What would we do in a case when
> > all
> > nodes shutdown at the same time due to a power loss (no ups)? Would the
> > cluster power up normally? Would we have to make sure to power up the node
> > that has control of the cluster group first? What if we didn't know at the
> > time which node had control of it?
> >
> > Thanks
> > Doug
> >
> > "Rodney R. Fournier [MVP]" wrote:
> >
> >> Oh yeah, forgot something. You may want to stagger the reboots. You can
> >> do
> >> this by modifying the boot.ini file.
> >>
> >> "Rodney R. Fournier [MVP]" <rod@xxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote in
> >> message news:eO8fg92TFHA.2756@xxxxxxxxxxxxxxxxxxxxxxx
> >> > Is that each node? That is not a lot of time. I hope you have power
> >> > monitoring software that can run a script and do automatic shutdowns.
> >> >
> >> > Cheers,
> >> >
> >> > Rod
> >> >
> >> > MVP - Windows Server - Clustering
> >> > http://www.nw-america.com - Clustering Website
> >> > http://www.msmvps.com/clustering - Blog
> >> >
> >> > "Doug" <Doug@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> >> > news:3731279B-6647-439D-8651-9CAE9B628699@xxxxxxxxxxxxxxxx
> >> >> John,
> >> >>
> >> >> On another note, we have a UPS that gives us 20-30 min of runtime in
> >> >> case
> >> >> of
> >> >> power failure.
> >> >>
> >> >> Thanks
> >> >> Doug
> >> >>
> >> >> "John Toner [MVP]" wrote:
> >> >>
> >> >>> Doug,
> >> >>>
> >> >>> The best way (IMO) would be to take all application groups offline,
> >> >>> then
> >> >>> shutdown servers individually...shutting down the owner of the
> >> >>> "Cluster
> >> >>> Group" last. You'd probably want to shut off your "passive" nodes
> >> >>> first,
> >> >>> then shut off all active nodes one at a time.
> >> >>>
> >> >>> BTW, I don't think you're going to have much control over this in the
> >> >>> event
> >> >>> of a power failure.
> >> >>>
> >> >>> Regards,
> >> >>> John
> >> >>>
> >> >>> "Doug" <Doug@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> >> >>> news:F4A5252B-C307-4607-AF80-A18B53E57851@xxxxxxxxxxxxxxxx
> >> >>> > We cannot seem to find documentation on how to shutdown an 8-node
> >> >>> > cluster
> >> >>> in
> >> >>> > case of emergency or power failure. Does anyone know where we may
> >> >>> > be
> >> >>> > able
> >> >>> to
> >> >>> > find this documentation or can anyone recommend what the proper
> >> >>> > shutdown
> >> >>> > sequence would be?
> >> >>> >
> >> >>> > Any help would be greatly appreciated.
> >> >>> >
> >> >>> > Thanks
> >> >>> > Doug
> >> >>> >
> >> >>>
> >> >>>
> >> >>>
> >> >
> >> >
> >>
> >>
> >>
>
>
>
.

Relevant Pages

  • Re: When a node fail I lost shared folders
    ... at the moment I can't power on the second node because the cluster could ... "Ramon Jiménez" wrote in message ... >> Fabrizio. ...
    (microsoft.public.windows.server.clustering)
  • Re: Unbloating the kernel, was: :mem=16MB laptop testing
    ... If you pay for electricity, ... Presumably a cluster exists to do more work than can be done on a single ... lot less power to do equal work. ... don't really have a good single term for an arbitrary unit of computing ...
    (Linux-Kernel)
  • SUMMARY: Redundant power supplies arent
    ... Sun replaced both power supplies, stating that both had failed at the same ... servers had incomplete prtdiags along with differing power packages.. ... cluster patches. ... ahead and rebuilt it using the jumpstart and then applying the cluster ...
    (SunManagers)
  • Re: Properly shutdown Windows 2003 Cluster during emergency power
    ... Losing power completely across the cluster could be very bad. ... you really want a node to come up and control the ... > shutdown the servers within 5 min of power loss. ...
    (microsoft.public.windows.server.clustering)
  • Re: Properly shutdown Windows 2003 Cluster during emergency power
    ... To add to Rod's commends...the first node that comes up will own ALL cluster ... wouldn't do this (I would rather control failovers myself) though you may ... > Losing power completely across the cluster could be very bad. ... >> shutdown the servers within 5 min of power loss. ...
    (microsoft.public.windows.server.clustering)