Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 07 May 2011 17:41:07 -0500
From:      =?ISO-8859-1?Q?Erik_N=F8rgaard?= <>
To:        Arun <>
Cc:        Lokadamus <>,
Subject:   Re: Link and network level in the tcp/ip stack
Message-ID:  <>
In-Reply-To: <>
References:  <>

Next in thread | Previous in thread | Raw E-Mail | Index | Archive | Help
On 7/5/11 4:12 PM, Arun wrote:

> Just add default route at your node, and make the default
> GW for this route as or the connected interface. Your
> SRV node should pass it to its default gw which should take
> care of forwarding it to the destination RN. If your SRV node could NOT
> forward the ping reply then add a specific route there like - "pkt comes
> from then forward it to
> Thanks.


There can only be one default gateway, anything else doesn't make sense. 
I did try adding a specific route on SRV for RN such that pings arriving 
on would be responded correctly. But, then RN can no 
longer reach No surprise there really.

So, why do we have this setup? Well, some services like ssh that is used 
for administration must arrive on 192.168.28/24 where as the commercial 
service has a dedicated network on 10.225.162/24 and to ensure 
availability and bandwidth we cannot accept to have ssh coming in on 
that network.

I should add that this is a Red Hat Linux, I ask here since the FBSD 
implementation of the tcp/ip stack is considered the reference 

So the question is which behaviour is correct, recommended or accepted? 
Stripping the link layer and reply according to the network layer, or 
keeping the link layer?

Thanks, Erik

Want to link to this message? Use this URL: <>