If DHCP and PDHCP are running on separate servers, check to make sure that the Tag 60 is NOT set in the global options for DHCP.
In a Novell Cluster Services environment, try these steps to resolve the symptom of the PDHCP server not responding:
1- Make sure that the PDHCP.NLM, PXESTART, etc. are not loading in the autoexec.ncf. They should load only from the NCS load scripts if they are intended to be a Cluster resource.
2- The PDHCP.INI should not be edited with IP addresses to bind the services to. The NLM does auto-discovery of the real and the resource IPs, and will work correctly if it is allowed to do so.
3- Typically, the TFTP directory should be on the shared, or virtual, SYS:\SYSTEM folder in the Cluster.
The best way to troubleshoot this error message is to perform a packet trace while the workstation is attempting to boot. What you should see in the trace is the workstation send a DHCP Discovery packet to the broadcast address. Then, there should be two DHCP Offer packets returned immediately to the workstation. One of them should be from DHCP with all the IP addressing information, Options 53, 54, 51, 1, 6, etc. The second DHCP Offer should be from the PDHCP server, and should include the Option 60, as well as some others, like Option 93, 94 and 97.
If the workstation sends 3 or 4 DHCP Discovery messages, and does not receive the PDHCP response it needs, then it will fail the PXE process and go on to the next boot device, typically the C: drive.