This is gonna sound dumb, but . .

Isn't the "master" node the one with the IP of the "cluster" object as
well? (meaning the little "molecule" NDS object in ConsoleOne actually
has an IP address and that's the "master" IP). The Master node always
resides on the highest numbered IP of the physical NCP server available
at the time.

The reason I ask, is this scenario:

"master IP" is:
4 servers running NCS have IP of:, 6, 7, 8 respectively.

What happens if, for whatever reason, during the install, somehow the
AFP/CIFS stuff is configured to bind to that master IP of

I'm not sure it's even possible, but maybe that would explain what
you're seeing?

We have a 16-node cluster and the master IP does kinda float around a
bit. The only time we've ever encountered issues with what server it's
on has been with the lovely backup software known as SyncSort.

In version 2.1.5d, they had an issue where the cmdnlm.nlm (I think
that's the name of it) wasn't been access correctly by BackupExpress.
BackupExpress uses the Master IP to query for the node location and for
some strange reason, on a full backup, it would abend the master node
(which would then fail to the next highest IP and it would promptly kill
that one as well, and the one after that, etc.)

After they re-coded the NLM it stopped doing this.


In article <oNaBe.4998$>, says...
> We don't need the master IP for GW, but the master IP will go to a node with
> GW. We like to keep GW on a node by itself. When the master IP goes to a
> GW Node, then we have to play the offline/online games till we get things
> back the way we need them.
> For some reason, our AFP does not work unless it is on the node with the
> Master IP, I believe our implementation of CIFS is the same.
> <> wrote in message
> news:%Y9Be.4901$
> >
> >
> >
> > "James Hevener" <> wrote in message
> > news:f58Be.4739$
> >> Thanks for the ideas on the Master IP order.
> >>
> >> "most everything to function ...."
> >>
> >> Two different answers on how the Master IP goes. One by highest IP, one

> > by
> >> order in which they come up in the cluster.
> >>
> >> Our situation is that we have some Mac users. We also have GW. GW is a
> >> high profiled app for us. With 4 nodes, the Master IP seems to jump all
> >> over the place - so I am guessing the guy who said it is order at which

> > they
> >> join the cluster. Yea, figure that one out and track it.
> >>
> >> Looks like our best solution may be to isolate GW into its own two or

> > three
> >> node cluster.
> >>
> >> We really like NCS. Instead of hours of down time, we are talking

> > minutes.
> >> But very high profiled minutes when it happens.
> >>
> >> Wouldn't you think in a release an idea would be to include a few

> > different
> >> options for setting the path of the Master IP???
> >>
> >>

> >
> > GW doesn't need the master ip address.
> > We have a two node gw cluster (this is because we are cheap not for
> > technical reasons) we have 2 post offices on it each on their own address
> > space they use thier own ip address and also different ports so there is
> > no
> > clashing when both post offices are on the same node.
> >
> > What's your issue with macs?
> >
> > have fun
> >
> > Marcus
> >
> >
> > --
> > Photos :
> > Blog :
> >
> > `The music business is a cruel and shallow money trench, a long plastic
> > hallway where thieves and pimps run free, and good men die like dogs.
> > There's also a negative side.' HST
> >
> >