At a customer site, I've shown that enabling PortFast resolved a DHCP
timeout problem being experienced with Zen7 imaging.

With PortFast DISabled:
- The PXE-DHCP request gets answered after quite a long delay - i.e. I
suppose just before it times out. But at least, it works.
- during suse linux stage the network driver is loaded and then another
DHCP request goes out. This one fails because the linux timeout is not
long enough. I assume during the network driver load, the PC NIC port is
reset and the Cisco STP cycle starts again (so need to wait up to 30 sec
for the port to wake up).

Obviously enabling PortFast speeds up the PXE-DHCP response and resolves
the linux DHCP timeout also.

However what if the customer does not want to enable PortFast on their
workstation-attached switch ports. Is there a way to extend the DHCP
timeout in the linux stage of the boot process? I suppose I may have to
modify something in the boot files? Or failing that, what strong
justification can I give to the customer that enabling PortFast is
generally a Good Thing (TM). I have already pointed them to the
STP/PortFast docs on Cisco site.

Graham Marsh