On Fri, 15 Jun 2012, Alberto MartÃnez wrote:
Our FR is doing EAP most of the time, and it's working fine. However, we would want our NAS to see the inner true User-Name, not the outer one. I know this can be set in the inner-tunnel post-auth section uncommenting the update outer.reply lines, but that exposes our users' inner User-Name to proxied-to-us authentications.
So my question is: Which attributes should I check to tell apart local and external auths?
We do exactly this and use Client-Shortname to control whether the inner username is revealed: we set Client-Shortname to be in the format '<server>@<domain>' - so we might have 'radius0@botolphs.cam.ac.uk' for server 'radius0.botolphs.cam.ac.uk', or 'roaming0@ja.net' for 'roaming0.ja.net'. This attribute is under our control with how we configure things in clients.conf, so we can ensure it isn't something bogus. We can then parse this with a regexp such as: if ("%{Client-Shortname}" =~ /\.cam\.ac\.uk$/) { # client is inside Cambridge - reveal the ID update reply { User-Name := "..." } } else { # client is not - set it back to the requested one (see note below) update reply { User-Name := "%{request:User-Name}" } } We do various other tricks with this - e.g. set Operator-Name depending on the college or department which forwards requests to us (since we act as a proxy to various groups inside the University, before passing things up to the UK national eduroam proxy service). Incidentally, on this matter, when I try and enable the relevant bit in inner-tunnel, post-auth - uncommenting: #update outer.reply { # User-Name = "%{request:User-Name}" #} ... I get a weird error when handling logins about the User-Name not matching (I haven't tried recently and haven't got something to test on). Instead, I use 'use_tunneled_reply = yes' in eap.conf and then use the above test in the default, post-auth section to reset it to something anonymous, if the client is outside Cambridge. I've never got to the bottom of why this is - I did search the archives and found other people with the same problem but never resolved it. - Bob -- Bob Franklin <rcf34@cam.ac.uk> +44 1223 748479 Network Division, University of Cambridge Computing Service