On Fri, 2005-01-07 at 09:20, Paul Stewart wrote: > Use *real* switches..;) Cisco all the way... Ummm, have you priced Cisco gig-capable ports compared to Dell... > I *believe* that Dell gets their switches from 3com but could be > mistaken... if that's true it would explain more... we rip 3com's out of > production all the time to replace with Cisco...:) No, it has to be SMC - the specs are identical. > > i am having a heck of a time getting a dell 5324 switch to dhcp > > directly to a couple linux boxes i have hung off it. > > > > enabling portfast on the cisco switches instantly fixed this problem, > > so i was suprised to see no change on the dell 5324. it's like i > > changed nothing at all. Just guessing: you probably have Ciscos on the same LAN and they are refusing to use the Rapid Spanning tree standard with other vendors, preferring to first attempt their proprietary PVST+. > > i have also disabled STP on the ports in question, but it still takes > > 1-7 seconds or more for the link to come up. this conflicts with a 5 > > second timeout in the network scripts. You really don't want to (and maybe can't) disable spanning tree completely on a switch. It is what keeps the network from melting when someone accidentally - or intentionally - makes a loop among the switches. The Cisco portfast setting just makes it forward packets while making the decision, which is still a little dangereous. > > for now we have upped the timeout to 15 seconds in the network scripts > > while we hunt for a real solution. Here is a link to a pdf describing how to make the Ciscos cooperate. I'd think it would make sense for a dhcp client to check for link before sending anything, but that's just me... -- Les Mikesell les@xxxxxxxxxxxxxxxx
