Hi, On 04/12/2020 18:55, Bryan Fields via 44Net wrote: This is perfectly fine and I'm not going to say you're wrong for doing it this way, but if you were building out something that could be used for emergency communication then it can pay to add authentication. You might also end up cornered into carrying potential personal information over your network (e.g. safety of life exemptions), and it'd be nice if you limited the scope of the interceptable traffic to only the RF portion of the network. In HamBSD (and OpenBSD) configuring IPSec for a GRE tunnel requires: 1. copy the public keys from one host to the other 2. add one line to /etc/iked.conf with the IP addresses and the names for the public keys you copied in step 1 3. enable/start/restart iked It's not magic, it's not mystery, it's actually pretty simple. That one line of config looks like: ikev2 'mb7uar_rsa' passive esp from 44.190.21.233/32 to 44.190.21.234/32 local 44.190.21.1 srcid hambsd.org dstid mb7uar.hambsd.org rsa This configures a tunnel that expects an inbound connection (MB7UAR has redundant internet connections, and is behind a NAT on both, so the other end is not fixed) with the first two IP addresses as the "internal" tunnel IP addresses. The third IP address is where to listen "outside" the tunnel. The srcid and dstid just match up the public keys to make sure you're authenticated. Thanks, Iain. -- https://hambsd.org/

In this case, why not use pptp? It does transport data over gre and offers a decent authentication using mschap2. If needed, one can also add mppe as encryption. I know there are theoreticians that state that it may be mathematical vulnerable because of "shortcut" implementations done by MS, but still no exploit is out in the wild. And the out-phasing of it in favor of strong cryptographic solutions actually lessens the pressure on really creating an usable exploit. On the other hand, ham radio networks are not an interesting target for potential crackers. On 04.12.2020 21:52, pete M via 44Net wrote:

The same firewall rules can be used to protect pptp servers, strengthening the system. But on the othe hand, the negotiation features of pptp offer dynamic routes on connect/disconnect, ip negotiation, subnet push and other goodies which are lacking in simple ipip or gre connections. On 04.12.2020 22:58, Jason McCormick wrote:

I never said it was "strong authentication", I said it "should suffice". I've been doing IT and security work for a long time so I chose my words carefully. For the actual threat posed to an amateur radio IPIP or GRE tunnel, the effort to deploy, maintain and troubleshoot complex technologies like IPSEC - especially by those who are not networking experts - is not justified by the actual threat posed to the connection. Could an adversary work out a man-in-the-middle attack, misroute your traffic, and spoof one end of your IPIP/GRE tunnel? Sure. Is it in any way likely to happen? Not in in the least. If you've attracted the attention of a malicious actor who is capable and willing to do that against your connection, then doing a 44Net connection to begin with is probably the wrong strategy for whatever the workload is and you have bigger problems. Increased security brings increased complexity. Yes, I can make a super-secure tunneled connection with PKI authentication, ephemeral keying, and all sorts of other things. But it's complex and largely unnecessary for the threat posed to some links used to cart around some digital voice traffic. Is an authenticated and private (encrypted) tunnel better? Sure. Is it necessary? That's a harder question. Jason -----Original Message----- From: 44Net <44net-bounces+jason=mfamily.org(a)mailman.ampr.org> On Behalf Of Iain R. Learmonth via 44Net Sent: Friday, December 4, 2020 4:11 PM To: 44net(a)mailman.ampr.org Cc: Iain R. Learmonth <irl(a)hambsd.org> Subject: Re: [44net] GRE tunnels Hi, I would have thought that this community more than others would understand that BGP (especially for AMPR resources) is not all that secure, and IP addresses are not strong authentication. It depends what you want to do though. If it's just for leisure and experimentation, then fine, you do not need encryption. If there are issues of abuse later, you can sort them out when they occur. For emergency communications, and also I guess because I come at amateur radio from a computing science background and not from a radio background, I prefer something a little more pro-active. I don't want to be finding that things have been randomly attacked (you don't need to be targeted to be attacked, look at ransomware it will just attack everything it can) mid-incident where you're trying to respond to some emergency and then your "resilient" communications turn out to be not so resilient. There is no right or wrong answer, it depends on your threat model, but you should give it some critical thought. There's a reason that most roles in large enterprises will have training on IT security even if the role isn't even remotely IT related. Thanks, Iain. On 04/12/2020 20:58, Jason McCormick via 44Net wrote: -- https://hambsd.org/ _________________________________________ 44Net mailing list 44Net(a)mailman.ampr.org https://mailman.ampr.org/mailman/listinfo/44net

Yes, there is a way. Just that you need to have that key exchange capture available. Which means access to a compromised system on the backbone or at the ISPs side, which also should allow you to do the capture. And that is note quite trivial and an effort not justified for cracking of a puny 44net gateway access with no added benefits. On the other hand, I understood that only specific implementations are actually vulnerable because of an implementation glitch where a key calculation is done just half the way offering a possibility to guess the password. Whatever, the ease of management of such a system outperforms a plain gre tunnel and discourages access of script kiddies. Of course, a serious hacker could probably do it, but to what end? I have such a system running for years, and apart from some sporadic login attempts with "root", "admin", "office", "sales" and such, and trying some dictionary passwords, nothing serious ever occurred. Just my personal opinion, Marius, YO2LOJ On 04.12.2020 23:13, Iain R. Learmonth via 44Net wrote: