Re: nodes will not shut down
From: Jog Dial (JogDial_at_discussions.microsoft.com)
Date: 01/18/05
- Next message: John E Davis: "RPC Unavaible remotely"
- Previous message: John Toner [MVP]: "Re: Three questions"
- In reply to: John Toner [MVP]: "Re: nodes will not shut down"
- Next in thread: John Toner [MVP]: "Re: nodes will not shut down"
- Reply: John Toner [MVP]: "Re: nodes will not shut down"
- Messages sorted by: [ date ] [ thread ]
Date: Tue, 18 Jan 2005 07:31:06 -0800
hmm, luck eh? After 25 years in the IT business supporting Solaris, Linux
clusters and every windows animal there is, I don't leave things to luck, I
test them until they are dead certs... I think it's a little cheeky to start
calling things lucky without all the facts.
I'm actually fairly sure I know what the problem is at this stage and as
soon as I get a quiet period, I'll be able to test it... not sure if anyone
is interested as everyone seems to think what I am doing is wrong, but if
this works, I'll be happy to let you know.
Cheers
"John Toner [MVP]" wrote:
> Jog,
>
> NT was a different animal...and if you had no downtime in an 8 year period,
> I believe that this had more to do with luck and the size of your business
> than your cluster configuration.
>
> In an ealier post, you said it would be possible to make another node a DC
> and move DNS to this node. This would be highly recommended, and also move
> all operational master roles to this non-clustered DC. If this were my
> environment, I would set it up with this external DC/DNS, and then perhaps
> use one of the cluster nodes (the passive or least taxed node) as a
> secondary DC. This way, your environment will still continue to function if
> you lose one server.
>
> You also mention that the private NIC didn't have a DNS record, and uses
> itself for DNS. You might want to read through the following KB for the
> recommended settings for the heartbeat NIC:
> http://support.microsoft.com/kb/258750
>
> Regards,
> John
>
> "Jog Dial" <JogDial@discussions.microsoft.com> wrote in message
> news:68308A72-59D1-47BD-AA17-93C5B3EFB48E@microsoft.com...
> > Ok, two severs, that's £1000 each, two more copies of windows 2003 server
> > £500 each, that's another £3000 pounds. UPS to keep them up, another
> £600.
> > That aint exactly peanuts - when I'm fairly sure there is a way to make
> what
> > I have work... just because nobody else has figured it out... like I say,
> I
> > had an NT cluster as PDC and BDC (which everyone fretted about) work
> > perfectly for over 8 years and the cluster never failed during that time
> and
> > that was before a lot of people even knew what clusters were. I guess I
> just
> > don't give up that easily and buy my way out of problems. I didn't have
> DNS
> > thrown into the equation I'll admit, but as I can see everyone here seems
> to
> > be loaded with more money than brains or will to figure things out, I'll
> > leave it at that and work it out myself. Don't get me wrong, I do
> appreciate
> > the advice, but if I can make it work as well as my last cluster, then my
> > company will be very happy and 0% downtime for a cluster over 8 years is
> > pretty track record in my view anyway.
> >
> > Thanks
> >
> >
> > "Russ Kaufmann" wrote:
> >
> > > "Jog Dial" <JogDial@discussions.microsoft.com> wrote in message
> > > news:F5E1C205-FA25-4112-AEB3-0FD6C6F14EC1@microsoft.com...
> > > > Hi,
> > > >
> > > > Thanks for the pointers, I can see a number of things to look for,
> however
> > > > I
> > > > can't agree with dumping the cluster nodes as DCs.s I only have one
> other
> > > > DC, If that goes down, I have no DCs. I've been running an NT cluster
> for
> > > > the last 6 years, one PDC, the other BDC and I never had a single
> problem
> > > > with it and I had no other DCs. So far, I've been very happy with
> 2003
> > > > e2k3
> > > > as cluster as well, but this problem has cropped up, recently as well,
> it
> > > > wasn't doing this to begin with... There is some form of DNS
> funkyness
> > > > going
> > > > on as it is registering the private addresses in DNS - the links you
> gave
> > > > speak of this and hopefully I can get that fixed as I think that the
> > > > private
> > > > network is the root cause of the problem.
> > > >
> > > > We run a pretty small shop here, currently only three servers on this
> > > > network, but I use a cluster as I need it to be up all the time, we
> only
> > > > have
> > > > about 50 users on it and the machines are way over spec in every way,
> and
> > > > I
> > > > just don't agree with not running DCs on them. That may be what's
> causing
> > > > my
> > > > problem in a roundabout way, hopefully geting the DNS straightened
> out
> > > > will
> > > > solve the problem - at least I've got quite a few things to try now.
> > >
> > > OK, I am confused. Uptime is extremely important to your company. Yet,
> you
> > > are running a configuration that is likely causing your problems. So, is
> > > uptime important or not? If so, buy two low-end servers for your DCs and
> get
> > > the DC functions off of your cluster. The low-end servers are not going
> to
> > > cost much at all. Small drive, single CPU, 512 MB of RAM, and you are
> ready
> > > to go.
> > >
> > >
> > >
>
>
>
- Next message: John E Davis: "RPC Unavaible remotely"
- Previous message: John Toner [MVP]: "Re: Three questions"
- In reply to: John Toner [MVP]: "Re: nodes will not shut down"
- Next in thread: John Toner [MVP]: "Re: nodes will not shut down"
- Reply: John Toner [MVP]: "Re: nodes will not shut down"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
