Hi,

On 03/01/2014 06:08 PM, Rob Janssen wrote:
> Arnold Schekkerman wrote:
>> On 03/01/2014 10:47 AM, Rob Janssen wrote:
>>> Ask Bjørn Hansen wrote:
>>>> On Fri, Feb 28, 2014 at 6:07 AM, Philip Gladstone <
>>>> [email protected]> wrote:
>>>>>>    Ok. I understand now. It looks as though the server would respond when
>>>>> the time wasn't known. It ought to say stratum 16, but maybe it doesn't.
>>>> It does, but it also says the time is off by 34 years.
>>>>
>>> Maybe it is a bug in the monitoring system.  stratum 16 means 
>>> unsynchronized, the
>> If I remember right, both + and - leap second flag set means 'no valid 
>> time', while
>> stratum 16 just means 'stratum 16' (which is usually configured as the local 
>> clock).
> I looked that up in the RFC before I wrote it.

OK.

>    Stratum (stratum): 8-bit integer representing the stratum, with
>    values defined in Figure 11.
>         | 16     | unsynchronized                                      |

Checking my memory: looking in the RFC's, curious if I remembered from an old 
spec.
and what the changes are. I found something tricky.

Searching in the (now obsolete) RFC 1305, for unsync only finds stuff related to
leap indicator (so far my memory is still fine :-) ).

Doing the same search in RFC 5905 gives an interesting result besides the table 
Rob
mentioned: section 11.1 "... then the leap is set to 3 (unsynchronized) and 
stratum
is set to MAXSTRAT (16). Remember that MAXSTRAT is mapped to zero in the
transmitted packet."

So, stratum 16 means 'unsync' internally, but is never transmitted over the 
wire!

The NTP RFC (present and past) is a tricky (confusing) specification as it uses 
the
same names for internals and for fields of NTP-packets sent over the wire (like
'mode' in section 3, which is even used wrong in figure 10 (no errata yet)).

To add to possible confusion, Figure 22 defines "Packet Type 6" *also* as
'unsynchronised'... I've not read further, so I am unsure if that is sent over 
the
wire or not.

>>> but it should not look at the actual time returned.
>> For the monitoring graph, that is no problem. Time graph and inclusion in 
>> the pool
>> are different things (though related of course).
>>
> I would say for plotting purposes it should be treated the same as "no 
> response at
> all".

It seems to me the monitor would never receive stratum 16, but Ask wrote it 
does...

Well, as long as the monitoring system rejects such servers from the pool, 
whatever
status indicator (or just 'plain wrong time') is used, it is fine :-)

Arnold
_______________________________________________
pool mailing list
[email protected]
http://lists.ntp.org/listinfo/pool

Reply via email to