UNIX admin writes: > One of the rare things I like is how Linux solved this: regardless > of the networking HW, all interfaces are named "eth[0-N]", for > example eth0, eth1, ... , ethN.
Linux isn't the only one to do this. AIX, BSD, and other variants do it as well. > Perhaps this would be a solution to consider? There can still be > bge0, and iprb0, iprb1 or ce0 in the system, but they would also > have aliases net0, net1, net2, net3 and so on. Yes, that's essentially one direction that flexible naming will allow us to drive. Frank Hofmann writes: > The problem with the Linux "solution" is that it breaks network/card > configuration assignments for network interface cards not soldered to > your motherboard. Which card in which slot/hotplug bay is eth0 ? Heck, we already have that problem in spades. Which one of my e1000gN interfaces corresponds to the one on the motherboard? Which to the ones on my adapter cards? How about the multiple ceX interfaces I've got? What happens when a card is replaced? Why should I have to hunt around to figure out that my motherboard has bge0 and bge1 but my add-in dual port card is e1000g0 and e1000g1? It's all pretty confused. I think we need a solid administrative solution here, and one that doesn't rely on obscure naming practices or "well known" device enumeration rules. > I also like the fact that for my laptop when running Linux, I don't have > to remember how the network driver is called. But the above problem, how > is it addressed on Linux ? On Solaris at least, that issue only occurs if > you have multiple cards of the same type... which isn't perfect either, > but in that scenario still better than linux. I disagree. I think it's baffling on both. -- James Carlson, KISS Network <[EMAIL PROTECTED]> Sun Microsystems / 1 Network Drive 71.232W Vox +1 781 442 2084 MS UBUR02-212 / Burlington MA 01803-2757 42.496N Fax +1 781 442 1677 _______________________________________________ opensolaris-discuss mailing list [email protected]
