On 29 Apr 2015, at 23:47, Sriram, Kotikalapudi <[email protected]> 
wrote:

> The above two BCP steps, if followed, will help prevent "couldn't validate 
> because of certificate lifetime".

Of course, most of the time this would happen correctly. People aren't stupid. 
But experience with HTTPS has shown that expired certificates can't be 
prevented 100%, and with routing the results are much more severe than with the 
web, as the user can't decide to go ahead anyway. The user wouldn't even know 
what happened. And fixing the problem may not be possible because of the 
unreachability caused by the problem.

> Second:
> The operational BCP can also say:
> Allow a certain grace period before you act on the update that became 'Not 
> Valid' due to cert expiry.
> (Earlier Sandy also mentioned this.)  

Ok. So what do you propose?

That a path with signatures based on expired certificates remains "valid" for 
some additional grace period? Then you simply postpone the problem by that 
period.

Or should operators have a way to differentiate between paths that are actually 
invalid and paths that are merely affected by expired certificates? In that 
case we are saying the same thing.

It might even be useful to have valid / expired-but-otherwise-valid / unknown / 
invalid rather than valid / unknown / invalid. Or maybe valid-algo1 / 
valid-algo2 / expired-but-otherwise-valid / unknown / invalid, so I can prefer 
paths protected with the strongest hashing algorithm over paths protected with 
a weaker algorithm and those over paths protected with expired certificates and 
those over paths not protected by BGPsec. (IMO invalid paths should not be 
allowed at all.)

Iljitsch
_______________________________________________
sidr mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sidr

Reply via email to