DNS & DHCP confusion

[CHR1S]

I have 2 domains admin and curriculum which run over the same physical network both completly independant.

There are no trusts and in theory no record of either on the opposing domain only the DNS entries further up the chain at the LEA could link them.

 

So anyway when I go to rebuild one of the RM PCs on the curriculum domain it assignes it an admin IP and the admin NS as its DNS but in the curriculum domain via the admin DHCP but it appears in the curriculum AD (you still following that? lol).

 

All non rebuilt RM PCs are fine and point to the right places and go pick up new IPs and what not when the leases expire or are deleted to test.

 

So yeh, how the hell is it doing that and why only when I us the RM build disks to reimage?

 

Thanks

 

Chris

[CHR1S]

Thanks for the reply jinn.

 

I have 2 scopes which up untill a week ago worked perfectly and have since I started here 18 months ago, all the existing PCs work fine and point to the right DHCP just not the ones that are rebuilt using the RMbuild disk which insists on picking the admin DHCP/DNS.

 

Ahh ill just wait for RM to figure it out haha

[SYNACK]

I have 2 domains admin and curriculum which run over the same physical network both completly independant.

There are no trusts and in theory no record of either on the opposing domain only the DNS entries further up the chain at the LEA could link them.

 

Your setup seems a little confusing, are your admin and curriculum domains segmented into distinct VLANs or do they all just share the same unsegmented local network. If they share then there must be something that distinguishes the two different types (curiculum/admin) machines apart so that the DHCP server scopes know which address to give them.

 

This is probably where the issue lies, if your rebuilt machines do not have the right flags specified in Windows (classid? - I'm not sure of the term for this) then it will assign the address from the wrong pool.

[CHR1S]

@SYNAK - Think you hit the nail on the head there! We have just had our core cisco switches replaced by the LEA my bet is the vlans they controll are not set up correctly anymore.

 

I think I may just merge the domains and have done with it!

 

Thanks for your help guys

[FragglePete]

That's the sort of setup I inherited when I started where I am now.

 

Two Domains but one physical network. The DHCP Server was essentially giving out IP Address to all machines. And for some reason the almost every device on the network had it's own reservation (I can understand part of the reasoning for that but it wasn't even structured!)

 

I've since tweaked things a bit, as although each DC had it's own DNS Server installed they were not talking to each other, so each DNS server now updates a secondary zone from each other. This has stopped a number of problems where seperate DNS entries were assigned to the same IP address. Was causing all sorts of havoc.

 

In an ideal world, managed switches with VLANs or seperate Network Infrastructures completely for each, but the infrastructure I've got and the number of unmanaged switches this just is not possible.

 

Because of the nature of DHCP Broadcasts, you can't have two DHCP Servers shouting across your infrastructure, unless each network is physically seperated or has routers in place that don't allow broadcasts (I think that's correct).

 

I'm planning on flattening the network in the not too distant future to make adminstration a whole lot easier, but for the time being I've got a trusted link between the Domains in place in preperation for Electronic Registration with SIMS. All very tightly locked down and controlled; We use CSE here so the Student's & Teachers machines are extremely limited to what they can and can't do, and for one they can't browse the network.

 

I don't have that many Admin machines anyway (no more that 25), so I've been moving these on to Fixed IP Addressing to ensure that they are pointing at the correct DNS Server. So far, everything is working sweetly that I've postponed the One Domain approach until maybe Easter next year.

 

HTH

 

Pete

[BrotherSidious]

You can put two DHCP servers on the same phsyical network BUT you can't predict which one will give out the address to the client. If you want to do this for redundancy you can set up an identical scope on each server. This ensures they will both give out addresses for the range you require.

 

The next step is vital though. On your first DCHP server (DHCP1) set up an exclusion such that it cannot give out the bottom 50% of your address range. On the second DHCP server (DHCP2) set up an exclusion for the top 50% of the addresses. With this system the same address cannot be offered by your servers (preventing conflicts) and if one is down the other can still offer valid addresses until it runs out. Clients will renew addresses from the DHCP server they initially get an address from so this can provide both redundancy and a bit of load balancing.