You are not logged in.
Hello ,
i want two connections to be active at the same time to my home little router (standars commercial lynksys model)
i have the eth1 interface and wlan0 interface
i have put a configuration files under /etc/network.d for each of them
when i connect by netcfg command , one of them alone , they both work , and i get the access to the internet
if i try to connect both , with the first everithing is fine . with the second i get from netcfg the following message:
RTNETLINK answers: File exists
> Adding gateway 192.168.0.1 failed
PS i am using the same gateway address on both the interfaces , because they aim to the same router
the configuration are both static IP
this is my first post i am a noob to Arch
Last edited by shadenzo (2011-05-27 04:41:29)
Offline
Yeah, bad idea... You'll end up with all sorts of asymmetric routing problems and confuse the poor little Linksys.
What are you trying to achieve by establishing parallel connections?
EDIT: wrong terminology.
Last edited by fukawi2 (2011-05-27 04:16:11)
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
hello
i want to achieve the most basic of the automation : when boot up connect with both interfaces , then if one is unavailable (e.g. ethernet unplugged ),
the other will be operational .
For the routing : i thought problems like that came from the network managers that try to automaticly establish connection whenever
sensing a change (for example the wireless becomes unavailable ecc)
If u know some specific routing problem , please explain me or put some link so i can learn about it
in short i want to do it because is the most basic automatism
Offline
Assume the following IP Addresses:
Ethernet: 192.168.0.111
Wireless: 192.168.0.222
Gateway: 192.168.0.1
Routing (your default gateway) is done by IP Address, so when you set your default gateway, how does the kernel decide to use the Ethernet or Wireless connection?
Solution is to set a dev argument to the default route:
ip route replace default gw 192.168.0.1 dev eth0
This forces the traffic for 192.168.0.1 via eth0 (ethernet). If the ethernet connection becomes unavailable, this can no longer work because you've told the kernel to use eth0 for 192.168.0.1.
Using route metrics also won't work.
Also, there's no guarantee which address a packet will be sourced with when it is sent from your computer. It could have the source .222 but be sent via the Ethernet connection. When the return packet comes, the gateway will send it to .222 via the wireless. This is confuse the kernel connection tracking.
It might seem like the "most basic automatism" but it's not. If you're on a mobile platform, I would suggest leaving Network Manager to manage the detection of a failed ethernet and switch to wireless, or use something like netcfg to manage the connection, but allow you to explicitly state which connection to use.
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline
hello
Ok seems i have to give up because it's impossible
well u can't have everything
thanks for the detailed reply fukawy2 : i'd like to learn more about routing kernel mechanisms (the basics)
if u have some link it would be appreciated
i consider the topic solved
(the terminology is not correct sorry but i am not english )
thanks
Offline
I don't have any links sorry... I learnt most of this information from a very wise and experienced guy I used to work with. We call him "Yoda". You get that when you're the guy who physically connected the first internet connection in Australia (many years ago)
Are you familiar with our Forum Rules, and How To Ask Questions The Smart Way?
BlueHackers // fscanary // resticctl
Offline