DNS issues setting up new forest alongside existing forest

[worcestertech]

What started off being a simple if complex task is turning into a stubborn set of problems.

 

In essence I am moving from an SBS2008 based domain to a server 2012 based VMWare domain.

 

The existing setup had to and has to be maintained until I have the new setup ready for a weekend switchover. In the meantime I needed to get the two forests talking to each other and both accessing the internet via the same gateway, an XTM505 on the existing 192.168 subnet with a static route set up to the 10.20 subnet.

 

Existing network is one SBS 2008 box and 5 other physical servers communicating over two flat HP 2610 switches. 192.168.1/24 subnet. DHCP and DNS have been disabled on this subnet.

 

New network one physical DC / DHCP server and 10 or so VMs. VMs running on two ESXi hosts with iSCSI SAN storage. The cluster communicates internally via two Dell Powerconnect switches. 10.20.30/22 subnet.

 

Interconnectivity is performed by a HP 3500 switch over a fibre link to one of the existing 2610 switches. The 3500 has two IP addresses, one on each subnet, these are the gateway addresses from and to each subnet.

 

Gateway addresses are 192.168.1.250 and 10.20.28.1 on the 3500 switch.

 

Firewall / Internet gateway is on 192.168.1.1 and is connected via a trusted interface to one of the 2610 switches which in turn is linked by fibre to the 3500 switch. Static route set up on the firewall as 192.168.1.250 which is an IP on the 3500 which in turn in connected to the 10.20 subnet.

 

DHCP server holds the above gateway addresses.

 

I've had the switch config checked out by a HP networking specialist who says all is well on the routing side.

 

Transitive forest trust has been set up from new forest to old. Trust cannot be set from old to new as SBS does not support trust (thanks Microsoft!)

 

DNS and DHCP are handled by the new DC on the 10.20 subnet. There are two DHCP scopes, on the 10.20 and on the 192.168 subnet.

 

I have two main issues that are proving stubborn to resolve:

 

1. Client PCs and servers on the "old" side are not seeing the DNS, and sometimes not the DHCP servers, i.e. they're not getting a response to the broadcast looking for DHCP.

 

2. The old gateway address of 192.168.1.1 appears to be persistent, ipconfig shows this address as one of the gateway addresses along with the new gateway. This address no longer exists in any DHCP server on the network yet if I carry out an ipconfig /release /flushdns /renew it reappears.

 

So, long preamble leading to a simple question: what am I doing wrong?

 

Does anyone have an idea of what steps I can take to resolve the situation?

 

Any suggestions will be gratefully received.

Edited February 17, 2014 by worcestertech

[worcestertech]

Bit more info that might help.

 

I've looked at the switch config in depth today. Added IP helper addresses and a static route / next hop address into each of the three switches. It seems that clients can now see the dhcp server but I still have DNS / gateway issues. Most clients still have no internet access.

 

Interestingly, if I point some clients at one of the 2610 switches address as the gateway internet access is restored. This doesn't work for all clients by any means though.

 

Any thoughts anyone?