[dns-wg] DNSSEC and DHCP

2023-05-22 Thread Julian Fölsch
Hi,

First of all: If you think, I should discuss this somewhere else, please tell 
me. :)

During my quest to get my SSH client to use SSHFP records and not annoy me 
with trust questions anymore, I fell into the rabbit hole that is DNSSEC.
Our domain already uses DNSSEC, so I only had to set up the resolver in our 
office and my PC to verify it.
This however had the side effect that child zones that are not signed were no 
longer resolving so I thought "Lets just sign them. Can't be that hard, 
right?"
I was very wrong.
One of the child zones is for hosts using DHCP and is managed by dnsmasq that 
unfortunately can't sign the zone.
But it can do zone transfers.
So we tried a setup using opendnssec as a signing proxy that transfers the 
zone to an unbound.
Unfortunately this has proven unreliable at best and broken at worst so I am 
looking to replace that.

I was just looking around for a DHCP server that directly can sign the zone 
but I was unable to find something so far.
So I was wondering how other people are doing this.

Are you signing DHCP zones?
Would you recommend (not) doing it?
If you are doing it, how are you doing it?


Kind regards,
Julian

PS: If you are at RIPE86 I also would be happy to discuss this in person :)


-- 
   Julian Fölsch
   
   Arbeitsgemeinschaft Dresdner Studentennetz (AG DSN)
   Teamsprecher Computing
   
   Tel.: +49 351 271816 69
   E-Mail: julian.foel...@agdsn.de
   
   StuRa der TU Dresden
   Helmholtzstr. 10
   01069 Dresden

signature.asc
Description: This is a digitally signed message part.
-- 

To unsubscribe from this mailing list, get a password reminder, or change your 
subscription options, please visit: 
https://lists.ripe.net/mailman/listinfo/dns-wg


Re: [dns-wg] DNSSEC and DHCP

2023-05-22 Thread Joe Abley
Op ma 22 mei , Julian Fölsch <[julian.foel...@agdsn.de](mailto:Op ma 22 mei  , 
Julian Fölsch < schreef:

> This however had the side effect that child zones that are not signed were no
> longer resolving so I thought "Lets just sign them. Can't be that hard,
> right?"

Verifiably-insecure delegations (a zone cut with no DS records on the parent 
side) should not be a problem to resolve through a validating resolver. You 
shouldn't have to sign your child zones to make them work. It seems possible 
that something else was wrong?

> I was very wrong.
> One of the child zones is for hosts using DHCP and is managed by dnsmasq that
> unfortunately can't sign the zone.
> But it can do zone transfers.
> So we tried a setup using opendnssec as a signing proxy that transfers the
> zone to an unbound.
> Unfortunately this has proven unreliable at best and broken at worst so I am
> looking to replace that.

There are a variety of other DNSSEC signers that can act as "bump in the wire" 
signers (where the "wire" is [AI]XFR). There are people who actually write that 
kind of software on this list and my hands-on with this stuff is a bit long in 
the tooth, so I won't try to speak for any of them.

> I was just looking around for a DHCP server that directly can sign the zone
> but I was unable to find something so far.
> So I was wondering how other people are doing this.
>
> Are you signing DHCP zones?
> Would you recommend (not) doing it?
> If you are doing it, how are you doing it?

It used to be quite common to glue DHCP servers to the DNS using dynamic 
updates, so that a DHCP server sends a DNS UPDATE when it wants to add or drop 
a binding to an address. If the DNS server handling the DNS UPDATE requests can 
also act as a DNSSEC signer, that might work for you. I have set up BIND9 like 
that before and it was fairly painless.

Joe-- 

To unsubscribe from this mailing list, get a password reminder, or change your 
subscription options, please visit: 
https://lists.ripe.net/mailman/listinfo/dns-wg


Re: [dns-wg] DNSSEC and DHCP

2023-05-22 Thread Joe Abley
Op ma 22 mei , Joe Abley <[jab...@strandkip.nl](mailto:Op ma 22 mei  , Joe 
Abley < schreef:

> Op ma 22 mei , Julian Fölsch <[julian.foel...@agdsn.de](mailto:Op ma 22 mei  
> , Julian Fölsch < schreef:
>
>> This however had the side effect that child zones that are not signed were no
>> longer resolving so I thought "Lets just sign them. Can't be that hard,
>> right?"
>
> Verifiably-insecure delegations (a zone cut with no DS records on the parent 
> side) should not be a problem to resolve through a validating resolver. You 
> shouldn't have to sign your child zones to make them work. It seems possible 
> that something else was wrong?

Actually, here's a thought -- check that the zone cuts actually exist (that the 
parent has a delegating NS set, and that the child has apex SOA and NS sets).

If your parent zone and child zones were hosted on the same servers, lack of 
zone cuts wouldn't matter if they were all unsigned (there's no referral to 
return, so the lack of a delegation goes unnoticed).

However you need the delegation to be present if you want to signal that the 
child zone is unsigned.

Just guessing, but I've seen this kind of thing before (and not just in 
enterprise zones).

Joe

>-- 

To unsubscribe from this mailing list, get a password reminder, or change your 
subscription options, please visit: 
https://lists.ripe.net/mailman/listinfo/dns-wg