Weird DHCP/VLAN behaviour

[GrahamL]

Hi all,

Some background first.

VLAN 100 - Data.  Windows SBS2011 (Windows 2008 Server) DHCP server, giving out ip in range 192.168.100.x

VLAN 200 - Voice, MITEL 3300 giving out ip in range 192.168.200.x to 5330 phones.

Option 125 set as below.  Option 43 also the same (but in hex)

id:ipphone.mitel.com;sw_tftp=192.168.200.253;call_srv=192.168.200.253;vlan=200;l2p=6;dscp=46v46s26

The issue :-

Since changing the Windows server from Windows 2003 to 2008, each morning, most computers that plug into the back of the 5330 phones pick up an ip from the MITEL unit instead of the windows server.  If I do a release/renew on the computer, it gets the correct ip from the windows server.

It's as if the phone is tagging the DHCP request from the computer with VLAN 200 tag, when it should not be tagging it at all.  This only happens when the computer is either first plugged in to the phone OR if the computer is plugged in but the phone is disconnected/reconnected to the switch/power cycled.

I know the switch tagging is correct.

The weird bit I can;t get my head around is that this did not happen on Windows 2003 and has only happened since the 2008 DHCP has been installed.  The option 43 has been added in order to try to fix this issue - I gather this is normal now anyway.

I can only think it is a bug in the phone firmware that is not now compatible 100% with the new Windows DHCP??  The phones say Main ver 01.05.00.41 from 2009.

Any one got any ideas on this?

THe only thing I can think of as a workaround is to set the "Client's class ID must match Name" to true.  Can you please review my two DHCP pics and see if I am coorect in the assumption that the DHCP "Name" on the MITEL must match the scope option of ipphone.mitel.com as shown in the Windows scope option 125?  Or am I looking at this wrong.

Thank you

Graham

[ralph]

I always ask this question when dealing with this type of DCHP issues because I've been bit a couple of times.

Is the data switch port connected to your DHCP server in Trunk mode or is it access only on the data vlan?
In your case is should be in access mode only on your data vlan.

Same question for your 3300.

Reason being, if both the DHCP server and the 3300 are responding to DHCP request, the 1st one to respond wins.

Ralph

[LoopyLou]

Don't see anything wrong on your programming. The funny part is that if the PC is getting an IP off the 3300 then it is not even talking to the Windows 8 DHCP server ( unless you defined the Voice VLAN on the W8 machine and both the 3300 and the W8 are giving out Voice VLAN IP's ).

How do you have the DHCP setup? Are your layer 2 switches using CDP or LLDP to advertise to the phones the proper voice VLAN so you don't need to use the double fetch method? As Ralph asks how are your ports setup?

 

[GrahamL]

I always ask this question when dealing with this type of DCHP issues because I've been bit a couple of times.

Is the data switch port connected to your DHCP server in Trunk mode or is it access only on the data vlan?
In your case is should be in access mode only on your data vlan.

Same question for your 3300.

Reason being, if both the DHCP server and the 3300 are responding to DHCP request, the 1st one to respond wins.

Ralph

Thanks Ralph,

Your post made me go back and check the VLAN trunking and tagging etc.  I discovered that the switch port for the 3300 was set to pass untagged packets on the data VLAN 100, which was incorrect.  Turning this off seems to have corrected the issue.

This issue has always been there on my network.  I just never saw it before and the change must be that the new Windows Server is just fractionally slower in responding to the DHCP requests than the previous Windows 2008 server.

CHeers

[sarond]

Normally if a Windows Server sees another DHCP server on the same subnet it will disable its own service because of a conflict.

[martyn]

It's funny how often this one comes up when someone has installed a new windows server and just plugged it in to a free port without checking the switch port config!

Took me a few hours to figure out the first time I came across someone that had done it, but you learn from that. 

[acejavelin]

It's funny how often this one comes up when someone has installed a new windows server and just plugged it in to a free port without checking the switch port config!

Took me a few hours to figure out the first time I came across someone that had done it, but you learn from that. 

Or when someone in a large enterprise network accidentally runs a patch cord between two switches, and the ports happen to be untagged on different VLANs... 

[GrahamL]

The thing that tricked me on this one was that the switch ports had been the same for three years and had previously never shown the fault sympton!.   The only thing that changed was the Windows Server from 2003 to 2008.  The switch ports had always been slightly wrong but this never caused the fault symptom until the Windows 2003 server was turned off.

We live and learn.

Thanks for the input guys.

Cheers

Graham