Weird dhcp issue last night

[John Blakley]

All,

We tried to move over our dhcp server to another server last night. The problem is that the server is on (10.10.10.50/24) and the core switch's vlan 10 primary is on 10.10.10.1/24. There are other secondary addresses on the same svi at 10.10.11.0/24 and 10.10.12.0/24. What we noticed was that anything in another vlan worked fine due to the helper statement on the svis, but anything that should have been on the 10.10.11.0 and 10.10.12.0 would send the broadcast, the server would see it, but the server would never send the dhcpoffer.

After calling Microsoft, it was determined that the reason for the issue was that the primary address for that vlan didn't have a scope (no 10.10.10.0/24 scope) to pull from so the dhcp server would never send an address from the existing superscope for the 10.10.11 or .12 subnet. They suggested that we put a relay agent on the switch, so for a test I configured the helper address on the svi. It worked, but I don't believe I'd want to keep it like this but can't find anything anywhere as to if we'd have problems with doing it this way. Keep in mind, the server is in the same vlan as the helper address is. I removed it and we reverted back to everything because my suggestion is that the systems department move the server to another vlan that doesn't have any dhcp scopes assigned to it so I can use a helper address across all user vlans.

In conclusion, this is what we ended up with and what kind of problems, if any, would we see? :

int vlan 10

ip address 10.10.10.1 255.255.255.0

ip address 10.10.11.1 255.255.255.0 second

ip address 10.10.12.1 255.255.255.0 second

ip helper-address 10.10.10.50

Server: 10.10.10.50

Scopes: 10.10.11.0/24 and 10.10.12.0/24

This works on the old server, but the difference is that the old server's primary address was 10.10.11.50 and running W2k3 whereas this is 2008R2.

Thanks!

John

[Richard Burts]

John

I am not sure why it works better with a helper address than it does when the server receives the original broadcast. But if it is working with the helper address then I would think that it is fairly safe to continue to use the helper address.

It may not seem logical to have a helper address configured for an address within the network on the SVI. But if it makes the server happy then I do not see a problem with it.

I have in the back of my mind a memory about something problematic about superscopes and DHCP on IOS routers. Perhaps it was something like this.

HTH

Rick

[John Blakley]

Thanks Rick My main concern is that the server possibly could be receiving 2 requests per host. I would think it could receive the broadcast from the host and unicast packet from the switch, so I'm not 100% certain if this would cause an issue or not.

Thanks!

John

[Richard Burts]

John

I would think that the server would receive two requests per host. It should receive the client broadcast and also the one forwarded by the helper address. If I understood your original post the server is not responding to the client broadcast request. So I do not see the second request as a problem.

The increased traffic on the network might not be desirable. But I do not believe that it is a real problem. And if two requests is what it takes for clients to get an address, then two requests would seem to be what you need to do.

HTH

Rick

[Rolf Fischer]

Hi John,

... because my suggestion is that the systems department move the server to another vlan that doesn't have any dhcp scopes assigned to it so I can use a helper address across all user vlans.

I think this would be the best solution.

You might alread know the "ip dhcp smart-relay" option, which can be (depending on your DHCP server configuration and capabilites) very useful when using secondary pools:

"If the ip dhcp smart-relay command is configured, the relay agent counts the number of times the client retries sending a request to the DHCP server when there is no DHCPOFFER message from the DHCP server. After three retries, the relay agent sets the gateway address to the secondary address. If the DHCP server still does not respond after three more retries, then the next secondary address is used as the gateway address."

http://www.cisco.com/en/US/docs/ios/12_4t/ip_addr/configuration/guide/htdhcpre.html#wp1089419

If you have smart-relay already configured, this could be the reason why it worked when you added the ip helper.

Best regards,

Rolf

[paul driver]

John,

Have you tried not using superscopes just use normal scoping?
How did you migrate the DHCP D/B ?

res

Paul