Hi Joe, Thanks for digging those up - the historical context is great. I hadn’t come across dns-flush before, and it’s interesting to see how the problem was approached at the time.
What strikes me most is that the operational need hasn’t really gone away. Once the problematic RRset(s) are updated, we want to consider the issue as fixed, but the cached state often prolongs the outage far beyond the actual time to repair. That gap between fixing the record(s) and full recovery of operations continues to be a recurring operational challenge.· EXPIRE is an attempt to address that long-standing gap in a standards-based way. Seeing prior work in this area, and learning of the feedback it received is genuinely useful in shaping the next revision. I appreciate you taking the time to share the background (and the humor). It’s good to be reminded that many ideas in DNS have deeper histories than I might have realized. Best regards, Duane > On Nov 21, 2025, at 09:46, Joe Abley <[email protected]> > wrote: > > Hi Duane, > > On 20 Nov 2025, at 17:30, Duane Powers <[email protected]> > wrote: > >> I have submitted a new individual draft proposing the EXPIRE opcode, >> which allows an authenticated authoritative operator to request >> immediate deletion of a specific RRset from a resolver cache. > > Allow me to dig into my historical outhouse of ideas, one moment... > > https://datatracker.ietf.org/doc/draft-jabley-dnsop-dns-flush/ > > The minutes of the dnsop meeting at IETF 87 contain recorded feedback on > draft-jabley-dnsop-dns-flush that was overwhelmingly negative, including the > phrase "the worst idea I have ever heard of" from Johan, I think with > reference to the general idea, not the proposed mechanism, and "I quite like > it" from Warren, so make of that what you will. I see that Warren and I > subsequently took the time to write up > https://datatracker.ietf.org/doc/draft-jabley-dnsop-flush-reqs/ but I think > we lost interest at some point after writing it. > > The slides I used included part of Randall Munroe's xkcd-1133 "you have a bad > problem and will not go to space today" without permission, under the title > "Requirements Analysis". Slides from IETF 87 no longer seem to be on-line, > which I think is a shame but which also at least avoids the obvious and very > reasonable copyright complaint. > > Your draft is surely better than mine, but I thought I'd take the opportunity > to bathe in the historical disgust of that moment once again. > > > Joe >
_______________________________________________ DNSOP mailing list -- [email protected] To unsubscribe send an email to [email protected]
