This is actually the config we use. We also pair it with a private CA to work around the OSes where CAT can only set a CA trust rather than CA and hostname pair.
There are a couple minor issues we've worked around: 1. A phone wouldn't send a separate outer identity. We also allowed a username@share-identity.<domain> outer identity to workaround. On the third RMA of the hardware it started using the outer identity properly, so I'm thinking this is actually some weird bug. 2. Windows uses the same @realm for inner and outer identity. Just be aware of this when picking the outer identity and documenting for users (our CAT page lists to use <username>@domain.edu for the Windows downloads). The only OSes we've seen as an issue have been the ones which can't support PEAP securely (see https://wiki.geant.org/x/MAB_AQ). For those the config fails anyways, and we recommend they use other options until they upgrade to something more modern (the problematic ones are long past vendor support at this point). On Tue, Jun 21, 2016 at 5:25 AM, Philippe Hanset <phan...@anyroam.net> wrote: > Curtis, > > Your comments made me think of a work around to make PEAP a little better > with CAT! > > Indeed EAP-TLS is by far the best way to avoid MiTM attacks, but for > institutions not willing to deal with EAP-TLS (cost of installer etc…), > Here is what one can do with CAT to promote the usage of the installer: > > In the CAT installer you can specify a fixed outer-identity (same for > everyone) either of anonymous@realm or *****@realm (***** being a long > string…but be careful some OSes do not accept this, but they all accept > anonymous) > You can then configure your home RADIUS server to only accept requests of > the form anonymous@realm or *****@realm and not accept username@realm. > > Users trying to configure manually will not succeed and will have to use > the CAT tool and be configured properly with a locked infrastructure > certificate. > > Some crafty people might end up guessing the outer identity (by sniffing > packets), but hopefully those ones are smart enough to know not to accept > evil twins RADIUS certs. > > This is not 100%, but it can definitely help! > > Philippe > www.eduroam.us > > > > On Jun 20, 2016, at 8:03 PM, Curtis K. Larsen <curtis.k.lar...@utah.edu> > wrote: > > > > The PEAP vulnerability is only mitigated by requiring EAP-TLS and > disabling PEAP. (It may help a > > little to recommend the CAT tool or similar, but not much) We've > recommended similar tools for 9 > > years - I know the take rates - they aren't great. Why? Because it is > optional. > > > > All I am pointing out is that one cannot say that they have completely > mitigated 100% the PEAP > > vulnerability while still running eduroam. I can say that for my > primary SSID. > > > > Thanks, > > > > Curtis > > > > > > On Mon, June 20, 2016 5:19 pm, Jeremy Mooney wrote: > >> How would you plan to mitigate for your users at remote institutions if > >> they're not verifying the certificate? It seems you can only prevent at > at > >> the IdP side of your radius infrastructure, and your clients can only > trust > >> they're talking to that server by verifying the certificate. If they > don't > >> verify the certificate, anyone can claim to be your server and just > allow > >> PEAP without you ever seeing the traffic. Technically that's also the > case > >> locally (someone else stands up an AP) and you could at most maybe see > it > >> happened but not block it (at least without going into the legal > minefield > >> of active rogue mitigation). > >> > >> I'd think that the best you can hope for (without solving the problem of > >> users falling for phishing/MitM in general) is just only allowing > EAP-TLS > >> so any client with a working config for your institution won't use PEAP, > >> but that doesn't require blocking PEAP on the SP side. > >> > >> > >> On Mon, Jun 20, 2016 at 5:00 PM, Curtis K. Larsen < > curtis.k.lar...@utah.edu> > >> wrote: > >> > >>> It's done on the RADIUS server, that's kind of my point. You have a > >>> service in your environment > >>> that may pose risk to some and you can't control it. > >>> > >>> I can mitigate the PEAP vulnerability for our users on campus, and our > >>> users at remote > >>> institutions, but I cannot mitigate that same vulnerability for another > >>> institutions' users on my > >>> campus. > >>> > >>> -Curtis > >>> > >>> > >>> On Mon, June 20, 2016 3:50 pm, Chuck Enfield wrote: > >>>> How would you disable PEAP on the eduroam SSID? I've never noticed a > >>>> setting for that. > >>>> > >>>> -----Original Message----- > >>>> From: The EDUCAUSE Wireless Issues Constituent Group Listserv > >>>> [mailto:WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU] On Behalf Of Curtis K. > >>> Larsen > >>>> Sent: Monday, June 20, 2016 5:19 PM > >>>> To: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU > >>>> Subject: Re: [WIRELESS-LAN] eduroam ssid > >>>> > >>>> Yes it does work. That's the problem - PEAP is vulnerable to Evil > Twin > >>>> attacks so we are disabling PEAP. Doing that on eduroam would break > all > >>>> institutions that still offer it. Leaving it enabled exposes users at > >>> our > >>>> institution. > >>>> > >>>> -Curtis > >>>> > >>>> ________________________________________ > >>>> From: Johnson, Neil M [neil-john...@uiowa.edu] > >>>> Sent: Monday, June 20, 2016 2:52 PM > >>>> To: Curtis K. Larsen > >>>> Cc: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU > >>>> Subject: Re: [WIRELESS-LAN] eduroam ssid > >>>> > >>>> eduroam should work with just about any authentication method that > uses > >>>> EAP (PEAP,TLS,TTLS) etc. > >>>> > >>>> So if your are say moving to TLS (Client certificates) it should still > >>>> just work. > >>>> > >>>> -Neil > >>>> > >>>> -- > >>>> Neil Johnson > >>>> Network Engineer > >>>> The University of Iowa > >>>> Phone: 319 384-0938 > >>>> Fax: 319 335-2951 > >>>> E-Mail: neil-john...@uiowa.edu > >>>> > >>>> > >>>> > >>>>> On Jun 17, 2016, at 10:19 AM, Curtis K. Larsen > >>>> <curtis.k.lar...@utah.edu> wrote: > >>>>> > >>>>> We're beginning to run into this problem as well. Luckily, eduroam > is > >>>>> not our primary SSID so at least the critical business functions > >>>>> continue to work fine on a separate SSID. My guess is that we'll > end up > >>>> turning eduroam off at those remote locations if problems get > reported. > >>>>> > >>>>> In talking with the eduroam admin from the other institution they > >>>>> mentioned that when this occurs in Europe the solution has been to > >>>>> change the name of the SSID. Is this really allowed? If so, I'm > >>>>> sold! Then we can start using our primary SSID with eduroam > >>>>> credentials! This is what I always thought eduroam should have been. > >>>>> To me the value was always in the universal credential > >>>>> *NOT* the SSID name. That was always a drawback for me especially as > >>>>> supplicants become easier to configure. > >>>>> > >>>>> The other problem that we're going to run into soon is that we will > be > >>>>> phasing out PEAP on our main SSID to mitigate against the evil twin > >>>>> vulnerability, but what do we do with eduroam? I mean I guess you > >>>>> could say it is the remote institution's problem, or the user's > >>>>> problem if they connect to an evil twin on your campus because > they're > >>>>> not validating the server. But if the evil twin is on your campus it > >>>> seems you have at least some responsibility in the matter. But as it > >>>> stands, eduroam will leave a bit of a gaping security hole for us. > >>>>> > >>>>> -- > >>>>> Curtis K. Larsen > >>>>> Senior Network Engineer > >>>>> University of Utah IT/CIS > >>>>> > >>>>> > >>>>> > >>>>> On Fri, June 17, 2016 7:35 am, Turner, Ryan H wrote: > >>>>>> Yes. We have a satellite school at UNC Asheville. Up until > >>>>>> recently, UNC Asheville was not running eduroam, and UNC Chapel Hill > >>>> was the only occupant of a couple of buildings on campus. > >>>>>> UNC Asheville adopted eduroam and wanted to move into adjoining > spaces. > >>>> So we were going to have > >>>>>> the situation where UNC Chapel Hill folks might attach to the wrong > >>>>>> institution's eduroam and vice versa. We ended up bridging the two > >>>>>> networks together through a single link, and based on realm, UNC > >>>>>> Asheville will terminate UNC Chapel Hill folks directly to our > >>>>>> network (through trunked vlans). It is nice, because now anywhere > on > >>>>>> UNC Asheville campus, UNC Chapel Hill folks have UNC Chapel Hill IP > >>>> space. Because it made sense, we actually turned off our access > points > >>>> and allowed UNC Asheville to provide wireless in our areas (so we > >>> wouldn't > >>>> have competing wireless). > >>>>>> > >>>>>> > >>>>>> 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 Becker, > >>>>>> Jason > >>>>>> Sent: Thursday, June 16, 2016 11:45 PM > >>>>>> To: WIRELESS-LAN@LISTSERV.EDUCAUSE.EDU > >>>>>> Subject: [WIRELESS-LAN] eduroam ssid > >>>>>> > >>>>>> Has anyone ran into this situation. > >>>>>> > >>>>>> We are an eduroam participating school and have multiple buildings > >>>>>> that are either across the road or sometimes sidewalk that another > >>>>>> University owns. The other school is wanting to join eduroam so my > >>>>>> issue is when we are both broadcasting the same ssid in possibly the > >>>>>> same airspace. I have a felling this is going to cause many > problems > >>>> as clients could bounce back and forth between systems. > >>>>>> > >>>>>> If you had to deal with this I like to hear your thoughts on it. > >>>>>> > >>>>>> -- > >>>>>> Thanks, > >>>>>> Jason Becker > >>>>>> Network Systems Engineer > >>>>>> Washington University in St. Louis > >>>>>> jbec...@wustl.edu<mailto:jbec...@wustl.edu> > >>>>>> 314-935-5006 > >>>>>> ********** Participation and subscription information for this > >>>>>> EDUCAUSE Constituent Group discussion list can be found at > >>>>>> > >>>> http://www.educause.edu/groups/< > >>> https://na01.safelinks.protection.outlook. > >>>> com/?url=http%3a%2f%2fwww.educause.edu > >>> %2fgroups%2f&data=01%7c01%7crhturner > >>>> %40email.unc.edu > >>> %7ccb70500b292d4427293208d39661db4b%7c58b3d54f16c942d3af08 > >>>> > 1fcabd095666%7c1&sdata=qGNRUEHsNMv7sMBIsc4xSekkNTdOESCI%2fPCz87RzRZY%3d>. > >>>>>> > >>>>>> ********** > >>>>>> Participation and subscription information for this EDUCAUSE > >>>>>> Constituent Group discussion list can be found at > >>>> http://www.educause.edu/groups/. > >>>>>> > >>>>>> > >>>>> > >>>>> ********** > >>>>> Participation and subscription information for this EDUCAUSE > Constituent > >>>> Group discussion list can be found at http://www.educause.edu/groups/ > . > >>>> > >>>> > >>>> ********** > >>>> Participation and subscription information for this EDUCAUSE > Constituent > >>>> Group discussion list can be found at http://www.educause.edu/groups/ > . > >>>> > >>>> ********** > >>>> Participation and subscription information for this EDUCAUSE > Constituent > >>> Group discussion list can > >>>> be found at http://www.educause.edu/groups/. > >>>> > >>> > >>> ********** > >>> Participation and subscription information for this EDUCAUSE > Constituent > >>> Group discussion list can be found at http://www.educause.edu/groups/. > >>> > >> > >> > >> > >> -- > >> Jeremy Mooney > >> ITS - Bethel University > >> > > > > ********** > > Participation and subscription information for this EDUCAUSE Constituent > Group discussion list can be found at http://www.educause.edu/groups/. > > ********** > Participation and subscription information for this EDUCAUSE Constituent > Group discussion list can be found at http://www.educause.edu/groups/. > -- Jeremy Mooney ITS - Bethel University ********** Participation and subscription information for this EDUCAUSE Constituent Group discussion list can be found at http://www.educause.edu/groups/.
