On Mon, 30 Jan 2006 10:07:26 -0600 John Haller <[EMAIL PROTECTED]> wrote:
> Stephen Hemminger wrote: > > > On Sat, 28 Jan 2006 20:31:18 -0800 > > "Ilya A. Volynets-Evenbakh" <[EMAIL PROTECTED]> wrote: > > > > > >>How does one enfrce eth device ordering at run time, when drivers are > >>compiled in? > One thing I have done is to build as modules everything that > I don't want to have the low numbered eth interfaces. > > > > Don't depend on the name! Use tools that use the MAC address. > > Several distro's do that, and you can but logical names on with nameif. > The one place that this does not work well is on the interface between the > boot monitor and the kernel, specifically the ip= boot parameter that > requires an ethN argument. I need this parameter because when NFS root > is enabled, without the parameter, the kernel will send DHCPs forever, > but the boot monitor has already identified the boot server in another way. > I generally don't use NFS, but I'm lazy, and don't want to build multiple > kernels. Then those bugs should be fixed. > Even at user level, MAC addresses for interface identification leaves > something to be desired, as it requires an administration conection > between a physical interface, which changes when a card is replaced. > This doesn't scale well, as less experienced administrators, such as > end users, have a hard time typing a MAC address, much less finding one. > Some kind of connection between the physical port location and the > logical name would work better here. A mapping between /sys/bus and > interface name assignment would be ideal for this. While generic > distributions would have a hard time with general mapping for any > generic platform, mappings for specific platforms could be generated. -- Stephen Hemminger <[EMAIL PROTECTED]> OSDL http://developer.osdl.org/~shemminger - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
