Qrux wrote:
> Bruce, Nathan, et al.,
>
> I've run into an interesting issue with our ifup/bridge/ipv4-static
> trinity-of-awesome, and it's closely related to Bruce's gut feeling
> that CHECK_LINK was...not quite right. Turns out, CHECK_LINK makes
> perfect sense--it was the bridge script that wasn't quite right...
>
> I ran into this because I needed to set MTU.
>
> CHECK_LINK is fine the way it's conceived. /sbin/if{up,down} should
> handle the PHY link, and issue the 'ip link' commands. So, they
> should bring the PHY link up and down. The problem was the wrench
> that bridging threw at it. Bridging sort of "inverted" the semantics
> of /lib/lsb/<service> by setting IFACE to the bridge interface. It
> inverted the semantics in ipv4-static. I just altered it so the
> semantics are identical.
I'm not seeing everything you are saying. Can you please send me a diff.
> Basically, in my ifconfig.brX (I'm using 1 in my example), IFACE
> needs to be the PHY link (eth1). That way, ifup/ifdown can treat it
> as a physical link. The beauty is that as long as "bridge" is
> specified in $SERVICE, it's perfectly clear which "interface/link" is
> the bridge:
>
> It's just ${1} passed from ifup/ifdown.
{$1} is the name of the config file extension. The design of the
scripts is to specifically separate the file extension from the
interface name. That is, you can easily have ifconfig.testbridge and
then use
ifup testbridge
and have the contents of ifconfig.testbridge, not the filename,
determine what to do.
I can see where you might want to set the MTU, but shouldn't any ethx
interface connected to the bridge have the same value as the bridge MTU?
-- Bruce
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
