24 Nov
2017
24 Nov
'17
10:37 p.m.
On Mon, 2017-11-20 at 15:49 +0100, Amuza wrote:
What is that WAN-to-WAN connection?
As I understand, for those devices that have a WAN socket (sometimes
also found as 'internet' socket), you connect them through the WAN
socket instead of the LAN socket.
You still need to config the static IP of the port... (longer
explanation, maybe for later!)
By local name resolution I mean something like a hosts file -to
be
edited manually- which is shared only within the layer-2 cloud.
There is no functionality like that.
Panos proposed an improvement to the current distributed DNS to be able
to define network aliases to the hostnames provided by the hosts (that
get registered by dnsmasq):
https://github.com/libremesh/lime-packages/issues/193
When you have different layer-2 clouds, I understand they
automagically connect to each other and share their Internet
connections.
You need to connect them through a wan-to-wan ethernet cable for
this
to happen.
Is that really necessary?
Not necessary, just a simple way to undestand it.
WAN: other networks
LAN: same network
I think I once tested it and it was not needed.
It is how it is recommended to be done, but maybe it works in other
ways...
I think the two layer-2 clouds saw each other
(wirelessly) and
automatically started to share the Internet connection.
Don't know...
But, what are the differences for users communicating
from
different
L2 clouds?
Different collision domains, no broadcast across different
networks.
I mean compared to users in just one L2 cloud.
They won't share the local name resolution anymore (published
services within the L2 cloud), will they?
Don't have any answer for this.
I have never checked it, but I think I read it
somewhere.
Maybe it is this A.L.F.R.E.D. that, besides sharing DHCP leases,
it
is also sharing the list of published services within the layer-2
cloud (pairing name and IP address)?
Only hostnames, no customizable network hosts file.
You can always flash a custom /etc/hosts file for your network