In <[🔎] 49DB5560.9000405@meetinghouse.net>, Miles Fidelman wrote:
>Thorny wrote:
>> On Mon, 06 Apr 2009 16:36:45 -0400, Miles Fidelman posted:
>>> Standard behavior, for years, has been to expect eth0 to be assigned to
>>> a machine's primary network interface. udev's behavior is more than a
>>> little counter-intuitive, and not well publicized or documented.
>> Well, it was covered pretty well in the release notes for Lenny.
>> Because I read them, it wasn't as surprising to me and I
>> found the documentation sufficient. That was my experience.
>That doesn't necessarily make it a good design.
Unlike /dev/by-{id,path,label,uuid} for disks, there's no equivalent of
network interface symlinks. I suppose the udev developers decided that
persistent device names were more important than "eth0" being the first
detected interface.
It should be noted that having "eth0" as the first detected device had it's
own set of problems. Users with more than one NIC often had them get
"random" names once the kernel's device discovery was made asynchronous.
Also, laptop users would sometimes have their NIC come up in the "wrong"
order because they booted with the PCMCIA/CardBus device already inserted.
Basically, the same issues that *still* cause problems when using
/dev/[hs]d* names for disks.
(The "real" solution is probably introducing the equivalent of symlinks for
network interfaces. That way, eth0 could be the first device the same way
/dev/sda is, but by-mac:011235facade would always be the same device just
like /dev/by-id/HITACHI_10K7_011235813 is. Patches welcome!)
--
Boyd Stephen Smith Jr. ,= ,-_-. =.
bss@iguanasuicide.net ((_/)o o(\_))
ICQ: 514984 YM/AIM: DaTwinkDaddy `-'(. .)`-'
http://iguanasuicide.net/ \_/
Attachment:
signature.asc
Description: This is a digitally signed message part.