Unexplainable Server - Sites and Services

[SG-W2K3]

Group,

I am very new to Active Directory and am trying to solve a mystery.

We have 20 offiices with WK23 DC'sin each office. Our Indy Data Center is
the Central Location with T1 links of varying speeds connected through the
Indy Hub(star topology).

In Sites and Services under NTDS Settings each individual office
consistently shows the same 3 Servers automatically generated by the KCC.
The first 2 in the list make sense as the are DC's located in the Indy Data
Center. The 3rd server however is a server in our smallest office (we'll
call it Madison) with the slowest connection (256 Frame Relay - most of our
others are 512 or higher).

This 3rd one appears in every Site and if you look at the Madison Site in
Sites and Services - under ntds settings - you will see every other server
in the organization listed as being automatically genereated.

We have deleted the manually and they have consistently re-appeared.

Does anyone have any idea why the KCC would be automatically generating this
and why it would keep showing up ?

I hope I explained that well and am looking forward to the responses.

Thanks,

Stuart

 

[Ryan Hanisco]

Stuart,

As I am sure you know, this is automatically being generated as the KCC
thinks that this is a replication partner with the other servers. I have
seen things like this happen in situations where there were overlapping
subnets, odd VLSMs used against your CIDR addresses, or servers that were
staged and moved between sites.

If you have multiple servers at some of your sites and a true star topology,
you may want to set up bridgehead servers for your sites and manually
configure replication partners -- its not difficult unless you have a mesh
network. Doing this will define traffic patterns and avoid the potential
for multiple replication paths through your frame cloud.

You can install the Support tools from the Server CD and use the REPLMON
tool to look at the replication patterns. This will give you a better idea
of how this is all working. DCDIAG also has switches that can help you look
at replication and configuration problems.

Perhaps Cary will have other ideas, but you'll want to avoid traffic taking
that extra route avoiding your core -- who needs an extra hop across a 256K
line?!

Let me know if you come across anything more specific or if you're looking
for better detail.

Ryan Hanisco
MCSE, MCDBA
Flagship Integration Services