Re: [Ntp] WGLC: draft-ietf-ntp-chronos

[Miroslav Lichvar <[e-mail]>]

On Tue, Aug 02, 2022 at 11:35:55PM +0300, Neta R S wrote:
> On Thu, Jul 28, 2022 at 5:38 PM Miroslav Lichvar <[e-mail]>
> wrote:

> > The document says Chronos uses the clock filter from RFC5905. That
> > requires at least 3 or 4 samples to get the first filtered sample. Is
> > it expected that the client will send multiple requests to each of
> > the randomly selected servers on the Chronos iteration?
> >
> Yes, this is correct.

Isn't that a waste of resources? Why not just skip the filter and work
with one sample per server? Chronos is not used for synchronization,
so I think it shouldn't matter if there is more jitter.

> > What should happen when there is no response from some of the servers?
> >
> The client will use the servers that responded.

Will not that make some attacks easier? Consider all the security
issues ntpd had where an off-path attacker could prevent the client
from accepting responses from any server.

I'd expect there to be a limit for the minimum number of responding
servers, e.g. 1/3 or 1/2 of poll servers, like NSANE in RFC5905.

-- 
Miroslav Lichvar