Skip to content

How should we handle upstream responses with TTL 0? #14

@emmaly

Description

@emmaly

Right now if something comes in with TTL 0, we don't cache it. But if there are multiple queued requests for the response, we give the response to all of them. Should we make a request unique for each request if we get a TTL 0? (How would we know until the first one came back? And what happens if the second response is not TTL 0?)

It feels like a TTL of 0 means that the result might be intended for a single recipient, not to be shared. Consider the case of a TTL=0 TXT record. Maybe that could be some sort of token or ticket that's intended to be used by only a single party, not to be shared.

Is there guidance, best practice, standard, advice on what to do with this?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions