search

ietf-wg-dnsop / draft-ietf-dnsop-glue-is-not-optional

1 stars 3 forks source link

Recommendations for when TC=1? #33

Closed wessels closed 2 years ago

wessels commented 2 years ago

When a name server sets TC=1 in a response the name server can either leave (a) the records that did fit, or (b) remove all those records and make a minimal sized response. Currently we don't say anything about leaving or removing records. Should the draft recommend one over the other?

ableyjoe commented 2 years ago

On Apr 15, 2022, at 14:45, Duane Wessels @.***> wrote:

When a name server sets TC=1 in a response the name server can either leave (a) the records that did fit,

RRSets, right? Not RRs. We wouldn't want to infer that it's ok to split RRSets to create a response that fits. or (b) remove all those records and make a minimal sized response. Currently we don't say anything about leaving or removing records. Should the draft recommend one over the other?

I don't think it's necessary to say anything about that. I think the adequacy of particular responses to particular queries (the subject of this draft) is orthogonal to the mechanics of truncating any particular response.

Joe