A few years ago I had somehow succeeded in getting WG adoption of 2 documents 
that addressed some pet peeves I had as a recursive DNS operator. Things got 
busy and my attention wandered elsewhere and I did not advance them. Since 
these issues continue to haunt RDNS operators, I have decided to update these 
documents. The first says that DNSSEC errors (and other auth RR issues) are the 
operational responsibility of and must be solved by auth DNS admins. The second 
says that people should not change to non-validating resolvers when a DNSSEC 
failure occurs. Both are likely obvious to us in the WG, but no so much to 
anyone else. ;-)

Just a week or so ago, Windows Update started to fail seemingly due to a bad 
delegation to a CDN from Microsoft and the TTL on the bad RR was long-ish 
(details are scant). So reporters and even Microsoft support started suggesting 
that people change their DNS resolvers. Only later did people figure out the 
problem was on Microsoft’s auth DNS end (see 
https://www.zdnet.com/article/windows-update-problems-fixed-now-but-heres-what-went-wrong-says-microsoft/
 and 1st story at 
https://www.zdnet.com/article/windows-10-updates-are-broken-again-but-this-time-its-not-microsofts-fault/).
 And we also see the issue of “DNSSEC validation failed, so switch to a 
non-validator” on a regular basis.

So I just submitted these again / updated them. I have asked the WG chairs to 
let me know how they’d like me to proceed with them, but haven’t yet heard 
back. In the meantime, I’m happy to continue to once again take input and 
comment from the WG.

https://datatracker.ietf.org/doc/draft-livingood-dnsop-dont-switch-resolvers/
https://datatracker.ietf.org/doc/draft-livingood-dnsop-auth-dnssec-mistakes/

Thanks!
Jason
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop

Reply via email to