[Fwd: [Debian-net-admin] autonet design]

[Jörg Sommer]

Hallo martin,

sorry for the late reply.

martin f krafft <madduck at debian.org> wrote:
> also sprach Jörg Sommer <joerg at alea.gnuu.de> [2007.06.08.1155 +0100]:
>> How does wpa_supplicant fit in the scheme?
>> 
>> layer 1. the daemon receives the event a new wlan interface is available
>> layer 2. the broker says it needs wpa_supplicant
>> layer 1. wpa_supplicant tells the interface is ready
>> layer 2. the broker says the interface needs dhcp
>> layer 3. dhcp_client is run
>> 
>> I think wpa_supplicant is somewhat special because it uses the same name
>> for the interface in different states.
>
> I am not sure exactly what you mean. Could you try to explain a bit
> more, please?

How does a second daemon fit into you concept? wpa_supplicant needs to
stay alive (it's not a one‐time task) to handle low level events like
disconnect from the access point or a request for rekeying. You must
continue listening to the daemon after the interface is up, because he
might tell you the connection is lost, while “ip link show” still says
the interface is UP.

I think about if wpa_supplicant is like ifplugd. It tells you the WLAN
cable is plugged in (the connection is established) and when the cable is
unplugged (the connection is lost). Yes, I think wpa_supplicant is
something like ifplugd.

Do you include support for ifplugd, i.e. do you start ifplugd when
ifup $IFACE is requested and wait for ifplugd says the cable is plugged in

And the second question I have, how do you propagate these events, the
cable is plugged in or unplugged to the DHCP provider? Do you kill him,
if the cable was unplugged?

Bye and good luck with netconf, Jörg.
-- 
Die meisten Menschen wollen lieber durch Lob ruiniert
als durch Kritik gerettet werden.