Many of our workstations are configured to PXE boot to a Ghost Multicast server. When we migrated our DHCP server from a Netware 5.1 server to a Netware 6 SP4 cluster, we found that when DHCPSRVR fails over to another server the workstations are unable to contact the Ghost server when rebooted. Using the -d3 switch and watching the packets we see the server receive a discover packet, send a reply, receive a request, and send an offer, but the workstation still won't boot.

But when we unload and reload DHCPSRVR on that same cluster node (the one to which it failed over), whatever the problem is gets cleared up and the workstation gets an IP address and boots from the Ghost server.

Anyone else heard of this?