Wednesday, May 01, 2013

Citrix PVS bootup failure with the Boot Device Management ISO

We had an issue recently with some Citrix Provisioning Servers (PVS) that were not getting a DHCP address when booting off of the Boot Device Management ISO (BDM.ISO).  What was happening is the server would just show this:

DHCP Discover ._._._._.

DHCP maximum number of retries reached.

When booting off of PXE or a WinPE CD I was getting DHCP without issues.

When I did a Wireshark of the line I saw the BDM.ISO boot send out DHCPDISCOVER packets but it never received a response.  I then Wiresharked the WinPE and PXE boots and saw the DHCPDISCOVER packet, followed by a DHCPOFFER, and so on.  When I examined the two packets I saw the BDM.ISO DHCPDISCOVER packet actually was a BOOTP unicast whereas the PXE and WinPE packets were BOOTP broadcast.  Thinking we had a DHCP Relay issue we checked our DHCP server (an InfoBlox server) and checked the logs for the MAC Address and here is what we saw when we booted with the BDM.ISO:

BDM.ISO DHCP traffic

The DHCP server was not responding to the DHCPDISCOVER.  This only occurred with the unicast packet and for some reason was "load balance to peer".  However it's setup, it appears UNICAST BOOTP packets are setup for load balancing but not sending a response.

PXE/WinPE DHCP traffic
The DHCP server is responding to a BROADCAST BOOTP packet in a very different way.  There is no load balancing going on and the server responds to the DISCOVER packet.

Unfortunately, we did not resolve this issue when I wrote this.  We got our farm to work by pointing the DHCP Relay to the previous DHCP server that is configured in such a way to resolve this DHCP request and present an DHCP OFFER.  Hopefully our network guys will get the DHCP fixed on the proper server.  If you are experiencing similar issues you may have a similar issue where the DHCP at your site is handling unicast DHCPDISCOVER packets differently then broadcast packets.

4 comments:

Parms said...

Hey - I am also hitting the same problem, did you ever find an answer?

Trentent said...

No. We just pointed at a different DHCP relay to resolve it. :(

Stan Svetec said...

Old post? I know. However, I just fixed this by giving my VM legacy NIC a new MAC address. Problem solved.

Trentent said...

Actually, eventually we did resolve this by updating our DHCP servers. Our DHCP servers were infoblox and updating them fixed this issue for us.