Bogus DHCP pool on fresh install

KOM

I've seen this twice now when playing in my home lab over the weekend.

WAN - 192.168.11.200/24
LAN - 10.10.0.1/24
DMZ - 172.16.1.1/24

During installation of pfSense, I told it to enable DHCP on LAN. When prompted for the IP range, I gave it 10.10.0.10 - 10.10.0.20. When I started up a client, pfSense gave it an IP of 192.168.0.100. When I checked my DHCP options, the default pool was 10.10.0.10 - 10.10.0.235. Once I set it to what I wanted and rebooted my client, all was well.

So, consistently it ignores the specified DHCP pool range that you tell it during installation and then gives an IP well outside that range.

Has anyone else seen this?

jimp

Did you set that on the console or in the setup wizard in the GUI?

KOM

Sorry, never saw your reply.

I set this on the console after install & reboot during the Set Interface(s) IP Address phase. I just set up two more nodes today for a lab environment and it definitely does it. The DHCP server does not respect the range you enter in the console. Today, for example, the specified range was 172.16.10.10 - 172.16.10.20. When you go into DHCP Server after WebGUI is alive, you see the range is 172.16.10.10 - 172.16.10.245. Every time. When I was using 192.168.10-.20, it came up as 192.168.10.10-.245.

phil.davis

That code uses gen_subnet_max() to find the range of subnet addresses valid for the DHCP pool. That function has a problem in pfSense 2.2 on 32-bit systems, and is fixed by:
https://github.com/pfsense/pfsense/commit/7094c303b7d46c9f7b24c3f1bd4432187832e85c

Something in checking/validation might be wrongly rejecting the DHCP range and thus not applying it.

Is this happening on a 32-bit install?

Can you apply the fix above, and see if it it works?

If not, then post the screen text of setting the IP address and DHCP range and I will see if I can reproduce it.

KOM

64-bit install running as a vm inside ESXi 5.5.0U2. The LAN & DMZ NICs are E1000 on private isolated subnets. I'll get some screens soon.

KOM

Here are two screens, one during install and one just after initial WebGUI config. DHCP range was setup as 10.10.5.10 - 10.10.5.20, yet WebGUI shows 10.10.5.10 - 10.10.5.245. This is in ESXi 5.5.0U2. I've also seen it in VMware Workstation and Virtualbox.

phil.davis

Do you get the question:
"Do you want to revert to HTTP as the webConfigurator protocol?"
Does it tell you good stuff like:
"Please wait while the changes are saved"
and so on?

The only way I can see it can bail out early is if you next say "y" to enable DHCPv6 and then press enter when it asks for the start of the DHCPv6 range - that will go straight back to the menu without saving anything.

All the platforms you mention are VMs. Do you have any idea if this also happens when running directly on hardware?
(I am struggling to see how VM vs real hardware will make a difference here!)

KOM

Yes, I go through the entire thing right to the end until I get the standard NICs & Options screen. One thing to add is that the LAN subnet already has a Windows AD DHCP server. He dishes out in the 10.10.2.x range. I don't believe he's part of the problem but I wanted to mention it.