> From my point of view the real criteria for removing the shaper device is
> a good document explaining how to set up the CBQ stuff to provide an
> equivalent setup. Once there is a good guide to transferring from one to the
> other then it is reasonable to drop shaper
I just started working on a Advanced-networking howto, which will
hopefully explain the new advanced networking completely enough. (Now that
I've spent countless hours reading the blasted code, fooling with it, and
using it in a production enviroment)..
The bigest problem I'm having:
I can't seem to manage to explain the new stuff (CBQ, 'fancy' routing,
qdiscs, etc) without throwing out the Unix compatible stuff (ifconfig, the
illusion of addresses bound to interfaces, etc) and just using the
iproute2 tools.
Are we going to depricate ifconfig? Once the new stuff is a bit more
polished will it be the official Linux userspace network config util set?
As you are aware, we no longer behave like a lot of other OSes in this
regard. On most systems, an 'interface' is a device with layer-3
addresses, etc.. In Linux an interface is now just a device, with queueing
params, and there are 'address' objects..
Ifconfig makes linux look like other OSes, but it isn't and this leads to
some confusion w/ non-obvious behavior..
I.e.
Hosta# ifconfig dummy0 1.2.3.4 up
Hostb# ping 1.2.3.4
1.2.3.4 is alive
Hosta# ifconfig dummy0 down
Hostb# ping 1.2.3.4
1.2.3.4 is alive
With the iproute2 utils, this behavior doesn't seem crazy:
Hosta# ip link set dummy0 up
Hosta# ip addr add 1.2.3.4/32 dev dummy0
Hostb# ping 1.2.3.4
1.2.3.4 is alive
Hosta# ip link set dummy0 down
Hostb# ping 1.2.3.4
1.2.3.4 is alive
[this makes sence because you turned off the link, but the packets
weren't coming in that link, so....]
Hosta# ip addr del 1.2.3.4/32
Hostb# ping 1.2.3.4
1.2.3.4 does not respond
There are a bunch of other situations where the ifconfig behavior leads to
confusing assumptions.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/