Dave Miner wrote: > Kyle McDonald wrote: >> Hi All, >> >> Currently I have the Solaris DHCP server setup to answer PXE boot >> requests and allow machines to JumpStart or KickStart. >> >> I recently installed a new switch to increase capacity (ports and >> backplane) and I can't be sure if it happenned at the same time but >> the PXE boots seem to be behaving differently now. >> >> On one set of machines that are directly connected to the same switch >> that the DHCP server is connected to, the Broadcom PXE boot banner >> spins for a while and prints 5 '.' before getting an answer. On >> another set of machines that is actually 1 switch away it takes 3 '.' >> before a response is seen. I remember this process being much quicker >> in the past. >> >> This is all with no DHCP relaying, and all on the same subnet. Snoop >> for some reason doesn't show a 'discover' request until late also and >> shows an answer for the first request seen, so if the PXE rom is >> sending out requests the whole time the spinner is spinning, then I >> don't see them coming through on the server. Where could they be >> getting lost and why? >> > > I would presume it's taking some time for link negotiation and STP. > Thanks Dave, and the other person who emailed me directly.
Enabling disabling STP on those ports by enabling 'portfast' solved the problem. Now I see 5 requests in snoop and debug output, 1 for each '.'. That's more like it! -Kyle > Dave
