Re: [Ntp] NTPv5 Loop Detection without Stratum - Why do we want this?

[Miroslav Lichvar <[e-mail]>]

On Tue, Sep 06, 2022 at 12:54:07PM +0200, [e-mail] wrote:
> > Von: "Miroslav Lichvar" <[e-mail]>
> > Stratum cannot prevent loops unless we don't allow clients to select a
> > server with a higher stratum (as it was in NTPv3).
> 
> Excuse me, I was naively assuming we were doing that (because otherwise, 
> why even use strata?).

I think we were doing that until NTPv4. In RFC1305 the test 7 checks
if the stratum is acceptable:
	pkt.stratum ≤ sys.stratum and pkt.stratum < NTP.MAXSTRATUM

I don't see that in RFC5905. Stratum is only included in the source
selection as an additional weight for sorting the sources. In current
implementations that weight is very small.

There is a paragraph indicating root dispersion was supposed to
prevent loops between peers at the same stratum:

   There is an important detail not shown.  The dispersion increment
   (p.epsilon + p.psi + PHI * (s.t - p.t) + |THETA|) is bounded from
   below by MINDISP.  In subnets with very fast processors and networks
   and very small delay and dispersion this forces a monotone-definite
   increase in s.rootdisp (EPSILON), which avoids loops between peers
   operating at the same stratum.

That doesn't seem to be working, because the client doesn't ignore
sources with a higher root dispersion and its own dispersion is based
only on the best source (not the highest dispersion from all of its
selected sources).

This is similar to the stratum issue.

At least this gives us a hint that even the NTPv4 authors thought that
loops should be avoided.

-- 
Miroslav Lichvar