"JDThree [MVP]" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> I didn't say that it was working fine - the problem is, as I stated, both
> cards active get me two registrations in WINS and DNS even though I've
tried
> to disable all services on the card that might have been involved with
> registering. This of course causes problems because clients often hit the
> wrong address and of course can't connect. The company has to keep going
into
> DNS and WINS and removing the "second" entry.
Ok I see. well, we may have to wait and see what other have to say about
that. I would have known to do only what you have already done.
> I've already got the unit connected through a crossthrough cable now
Good, the cross-over cable is the best way to keep that portion of it. Don't
let anyone talk you into changing that.
> And talking on both subnets - yes, the RAID *DOES* exist on both
subnets...
> The management port on the device that's set for the "bogus" subnet has to
be
> in communication with the server - the serveraid program controls the unit
> via that second adapter and non-standard subnet, while the ISCSI initiator
> actually uses the common IP subnet connection to the main switch to
actually
> *write the data* that's going to and from the drive. So yes, it makes
perfect
> sense. This is ISCSI, and that's how IBM has forced this particular unit
to
> work. The ISCSI initiator on the server's "normal" network cannot be in
the
> same subnet as the management port that the server uses to control the
> device.
>
> Sorry to have asked the question - I didn't realize that ISCSI was going
to
> be such an unknown factor here. Should have known since even IBM had
problems
> getting the unit up and running because it's such a new device.
Relax, you are talking about a network that you can see, feel, and touch
and you know how you built it. While I am talking about a network I have
never seen and have no background knowledge of and have no information on
other than a few lines in an email about a device I have never heard of. so
you will have to allow me a little "catch-up time".
Network Engine's NS6300 Firewall works just like you describe. It has a LAN
interface and a management interface (different subnets). The management
interface connects to a second Nic in a "management" computer. It works
perfectly fine, however the management computer is not a DC running DNS/WINS
and "routing" is not enabled, so the rest of the network never even sees the
management segment and nothing ever gets registered,...so all is happy.
Once the initial setup is done it can be configured to allow management on
any other port so the "management port" is no longer needed and can be
disconnected.
I don't really see a way around your problem if the DC is going to be *both*
the machine that actually uses the Array as a logical drive,..and as the
management machine at the same time. how about finding an innocent
workstation to function as the management machine and have the two nics in
it,..that would solve the whole registration issue. In similar situations I
just have a laptop that I carry around as a "management device" and just
temporarily hook it up to things like that when I need to change
something.What does IBM have to say?,...surely they have run into this issue
and already have a solution developed for it.
--
Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com