Joe Smith napisał(a):
Perhaps, if squid does get confused (two
squids on one IP address) you could run the second instance on a fake IP
address/alias but I don't know much about that.
And I know all to much about that. IT is done by creating a new loopback
ethernet device, and briging it to your real device. to get an ip addess
for this device you send out a dhcp broadcast on it, which loops back,
crosses the bridge and enters the real network. (unless your computer is
the DHCP server). The exact same thing happens with outbound packets.
Inbound packets might not need to cross the brige and loop back. I think
linux is smart enough to realize that the addressed ip is the same ip as
a device to which it is bridged, and tus insert it directly in the
virtual device's incomming queue.
A lot of work, but certainly intersting.
[Unfortuantely this knowlege is Windows based, and linux may need some
coaxing to accept it]
There ain't no such thing in linux as "loopback ethernet device". You
can have either one or the other.
If you want to learn a little about "routing with loops", try my article
on netfilter mailing list.
https://lists.netfilter.org/pipermail/netfilter/2005-April/059970.html
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
