Stephanie Daugherty <sdaughe...@gmail.com> writes: > Having dealt with systems with half a dozen interfaces in the past there's > a very small number of cases where it even matters, but when it does, it > can be a huge inconvenience either way.
Namely, "it matters" because Dell boxes with more than one interface usually come with a mulitport NIC and a 'randomly chosen' additional one often/ usually requiring more than one driver. That's easy to deal with for the people using these Dell boxes (compile the drivers into the kernel) and certainly not a reason why anybody else should suffer. > Their (for once, rational) argument is that Interface names of existing > interfaces should never change by adding or removing hardware from the > system, Or, putting this to a real background: Suppose your old eth0 fails and you put in a replacement, you then have to change everything over to eth1 because you get a new name for that (instead of the much more sensible idea that 'eth0' is the first interface found by the kernel) > or as a result of hardware/software quirks (on some systems, > interfaces are known to swap places on reboot, which is clearly not > acceptable). Have you considered using screws to fix them to their slots if they're that jumpy? There's no "hardware quirk" which causes PCI addresses to change[*] and the 'software quirk' is called udev. [*] As far as I know. _______________________________________________ Dng mailing list Dng@lists.dyne.org https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
