[Dxspider-support] Secure node to node connection proposal

Dirk Koopman djk at tobit.co.uk
Sun Feb 26 16:54:18 GMT 2023


I must agree with Steve that mandating SSL/HTTPS for internode or user 
connections is a step too far and is a solution for a problem that does 
not really exist. And, even if it starts to exist in the future, there 
is enough hardening in the code already to deal with it (yes, yes I know 
- hostages, fortunes etc). But, in a separate email, I have made a 
suggestion as to how it *could* be done - if someone cares to do the 
work and contribute that to the "product".

But I would have to add that, even if SSL connections were implemented, 
I disagree profoundly that explicitly sharing registrations and 
passwords through the network (whether encrypted or not) is a good idea.

If users want to use the same password on two or three nodes, I want 
them to do that themselves. Making users register in the way outlined 
below provides just enough friction to prevent random "node tourism".  
What controls are going to be effective if some bad actor (finally) 
registers on one node, has his (pronoun deliberately chosen) credentials 
carefully (and securely) transferred to every other node - which then 
allows him to wonder around all the nodes at will, spreading ordure as 
he goes?

So I am for all users to register on each node that they use AND 
(probably worse from the ungodly user's point of view) have the node 
check that they use the same details as that node may already have 
acquired (some of which will not be displayable to unprivileged users) 
through PC41 or, now increasingly likely PC91 'all in one' sentence.

To answer Steve's suggestions below, I think could implement this and it 
is a good way to go. It is on the TODO list.

Thank you Steve.

73 Dirk G1TLH

On 26/02/2023 11:11, Stephen Carroll via Dxspider-support wrote:
> I applaud the effort put forth with this detailed proposal. However, 
> it's implementation would rival the military encryption protocols I've 
> dealt with for the past 40 years. We're not trying to pass nuclear 
> launch codes on a highly secure network!
>
> Yes, these proposed steps will greatly change the entire cluster 
> system. It will basically mean a majority of worldwide nodes will go 
> QRT. That should NOT be the goal of anyone, since without nodes, users 
> and DX spots we have no network.
>
> REGISTRATION
>
> I do think, at a minimum, ALL nodes should require registration... and 
> it should be the default (or hard-coded) in the software. However, I 
> suggest the registration process be completed entirely without any 
> SysOp action. When a new user logs into the cluster, it asks for Name, 
> QTH, QRA and Homenode. I would add Email to that list and change to 
> HomeQTH and HomeQRA as well. Once a user successfully enters this 
> basic information they are registered on that node. This puts the 
> registration process in the user's hands! In fact, maybe the system 
> could lookup via QRZ.com to ensure its a valid callsign, QTH and QRA. 
> Current node users should be required to enter the same information to 
> complete the registration process as well (This requires users to 
> actually view the telnet login window on startup though, which many 
> don't even use anymore). We could then use the MOTD and MOTD_N files 
> to instruct and guide the user accordingly.
>
> PASSWORDS
>
> If enabled in /spider/scripts/startup, the node would automatically 
> generate a temporary password for the user upon login. Then the user 
> would be required to create their own personal password:
>
> set/password <abc123>
>
> Again, no SysOp input required! In fact, the initial temporary 
> password could be the same for ALL users to make it easier.
>
> Just a few random thoughts...
>
> 73, Steve - AA4U
>
>




More information about the Dxspider-support mailing list