I'm currently working from a customer LAN which is "protected" through authentication through captive-portal-like technology.
In theory, first thing you do in the morning is click on some URL, get the loging form spoofed by the captive portal and authenticate to the network (and then get on with work and other non-transparent-proxy frustrations).
Of course all these 'clever' captive portal techniques (read MITM attack) are easily detected by today's browsers and most of the time, the login form never shows up and all you get is a timeout.
Browsers occasionaly report self-signed certificates or bad SSL packet (TLS 1.0 I guess). But it also looks like my company AV (Symantec) is not even giving browsers a chance.
I wonder if I could pass this hurdle with some kind of older version of some browser I'd use only for the purpose of login in. Surely I'm not the only one stumbling on this issue. Any suggestion?
Thanks all for all suggestions (+1'd). Here is the progress so far.
The VPN is Cisco AnyConnect. There is no difficulty accessing the VPN from its public (Internet) leg. I'm using OpenConnect on Mint. All OK but slooow.
When on the faster company LAN, however, several hurdles come into play. Access to the LAN is guarded by a Cisco Product named ISE (Identity Services Engine) version 1.0. From a user's perspective, this ISE works like a captive portal, spoofs DNS requests, redirects to a portal that authenticates (radius) and apparently also uses TLSv1. Browsers are unimpressed. OpenConnect also complains (can I tell OC to be more leniant?). Also OCSP times out because servers can't be reached.
I'm planning to script all that stuff with python and
requests to go through all steps in a controlled sequence and check each step's execution. I initially thought capturing 'curled' requests in Google Chrome developer tools would suffice, but the infra seems very shaky and python+requests looks more promissing than curl+bash. To spice things up a bit, there is non transparent proxy.