So, for the time being, I am adding some very simple javascript to the landing 
page.  If it detects Android 8, it will display a message telling the user that 
after login, they must close the browser and open a new one and goto 
wifi.unc.edu.  I am going to play around with the browser environment variables 
and see if it is possible to discover if they are in the pseudo browser (look 
at the difference in environment variables between the full browser and the 
pseudo browser).  If so, I can just take away the login option until they open 
a browser with full power...

From: Turner, Ryan H
Sent: Wednesday, September 6, 2017 9:54 AM
To: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU
Subject: RE: Defeating Android 8.X Captive Portal detection

We haven't had the problem with OSX.  I worked hard to get rid of captive 
portal detection on all browsers.  Everything has been great, until now.

We have a setup like this:

We use a pfSense firewall on an onboarding SSID.  Users have 2 states:  
unauthenticated and authenticated.  Prior to be authorized (which requires 
logging in on a web portal), their connection is extremely limited except with 
whatever holes I have poked through to defeat captive portal detection and make 
things smoother.  One they are authenticated, there is NO ACL on the back-end, 
but I do blackhole a bunch of popular sites (through DNS redirection) so people 
will not use the onboarding SSID for browsing.  For the new Android, with our 
setup, the pseudo browser remains open, even post authentication (which 
probably means one of those black holes sites I have is being checked and still 
doesn't indicate connectivity).

My 'workaround' for the moment is to allow google.com to go through...  which 
is not a good one.  Since google.com is most people's default page, it means 
they will NOT get a wireless redirect to login and authenticate until they 
browse somewhere else.  So I don't think I am going to keep this.  I may just 
have to add some verbage to the login page that educate android users, but it 
will probably only be read 1 out of 10 times.

I really hate how it feels like we are having to constantly work against google 
with this stuff...


Ryan Turner
Manager of Network Operations
ITS Communication Technologies
The University of North Carolina at Chapel Hill

r...@unc.edu<mailto:r...@unc.edu>
+1 919 445 0113 Office
+1 919 274 7926 Mobile



From: The EDUCAUSE Wireless Issues Constituent Group Listserv 
[mailto:WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU] On Behalf Of Wyatt Schill
Sent: Tuesday, September 5, 2017 4:53 PM
To: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU
Subject: Re: [WIRELESS-LAN] Defeating Android 8.X Captive Portal detection

This is the same problem we have with Mac laptops, the 'pseudo' browser will 
allow the user to run through the whole onboarding process until the final 
download step where it refuses to allow the user to download a config file.

Our only fixes are to continually educate users to close the 'pseudo' browser 
and open a full version of safari, or to add pre-auth acls to allow the device 
to fully access the apple urls it is checking so that the 'pseudo' browser 
never pops up and the user manually opens a standard browser to get the initial 
captive portal.

Looks like android will need something similar.  (although we already have some 
of google open to allow guests to use google credentials to authenticate, so it 
probably won't be much extra to add)

Wyatt Schill
Senior Network Engineer
CCNA-Security, CCNP-R&S
Green River College
12401 SE 320th St. Auburn, WA 98092
wsch...@greenriver.edu<mailto:wsch...@greenriver.edu>
[Green River new official mascot logoEmail]

From: The EDUCAUSE Wireless Issues Constituent Group Listserv 
[mailto:WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU] On Behalf Of Turner, Ryan H
Sent: Tuesday, September 5, 2017 1:34 PM
To: 
WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU<mailto:WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU>
Subject: [WIRELESS-LAN] Defeating Android 8.X Captive Portal detection

Even though Android is only 7% of our install base, it amounts to 75% of my 
problems...

It 'appears' on first glance that google has changed the captive portal 
detection on version 8.  It 'appears' (very early into this, so this may 
change) that google now checks for both a generate_204 on both 
connectivitycheck.gstatic.com and a gen_204 on 
www.google.com<http://www.google.com>.  Why is this a problem?

We, as many people do, have a onboarding SSID.  TLS requires proper onboarding. 
  That means that we need to process people through the portal in an orderly 
manner to get them where they need to go.   When Android 8.X detects a captive 
portal, it will prompt the user to 'sign in'.  This process opens a pseudo 
browser (a browser that is limited in what it can do) to the captive portal 
login.  After the user logs in, the user stays inside of the 'pseudo' browser.  
The browser has limited powers, and apparently will not allow the user to 
download or install an agent or configuration files.  You can see the 
problem...  They will get to the onboarding page, and nothing will work.

I've managed to 'by pass' the problem, but it isn't ideal.  Has anyone else 
seen this with commercial portals and figured out ways around it?

It is possible all of this is in error, but I just got done with a bunch of 
packet captures that seems to validate this.  I only have one Oreo user in my 
vicinity, so I will need to get my hands on a few more to see if this really is 
an issue, or just bad luck.

Ryan Turner
Manager of Network Operations
ITS Communication Technologies
The University of North Carolina at Chapel Hill

r...@unc.edu<mailto:r...@unc.edu>
+1 919 445 0113 Office
+1 919 274 7926 Mobile

********** Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/discuss.
********** Participation and subscription information for this EDUCAUSE 
Constituent Group discussion list can be found at 
http://www.educause.edu/discuss.

**********
Participation and subscription information for this EDUCAUSE Constituent Group 
discussion list can be found at http://www.educause.edu/discuss.

Reply via email to