Re: [Trans] Gossip: Unsticking a client caught with potential evidence of log misbehavior

On Tue, 20 Oct 2015 at 16:28 Tom Ritter <tom@ritter.vg> wrote:

> On 20 October 2015 at 08:46, Eran Messeri <eranm@google.com> wrote:
> > +1 for reporting it backed to the alleged origin - the problem Tom
> describes
> > is significant as the client would want to hold onto these suspicious
> pieces
> > of information until it is confident it has been reported, but has no
> way of
> > knowing if the report has been acknowledged by the 'genuine' origin.
> >
> > One option is to not prescribe when clients drop such pieces of
> information,
> > so clients are free to re-send it multiple times or even forever, until
> an
> > out-of-band signal (i.e. a client software update) is received to
> indicate
> > to the client that this information has been received and acted upon.
>
> So reporting it back to the origin is an appropriate solution for a
> subset of situations.  If I have an SCT+certificate I want to resolve
> to a STH, but can't, I can report it up to the origin via SCT
> Feedback.  We also (or will) have guidelines for submitting it
> multiple times, and ensuring that it's reported on a connection that
> 'looks okay' (we have some notes in 7.1.1)
>
> There are (at least) two cases that aren't covered by "report it to
> the origin". (Maybe more, but I can only think of two right now.)
>
> - The origin doesn't implement SCT Feedback.
>

If it turns out that feedback is essential and the origin doesn't implement
feedback, in the medium term we can assume it doesn't care about CT.

- The data I have is an older STH, which is not tied to any specific origin.
>

Then why do you care? Can't you freely gossip the STH?

>
> -tom
>
> _______________________________________________
> Trans mailing list
> Trans@ietf.org
> https://www.ietf.org/mailman/listinfo/trans
>