[ntp:hackers] GPS weirdness with ntp-dev-4.2.0a-200503xx
John Pettitt
jpp at cloudview.com
Sat Apr 16 18:09:57 PDT 2005
Mark Martinec wrote:
>Perhaps the question is:
>- have the specs (rfc1305) changed in the beginning of March,
>- or, was previous implementation incorrect,
>- or, is the implementation now incorrect.
>
>Possibly an important note: several (most?) refclocks mentioned in
>problem reports seem to be without a PPS signal / atom driver.
>Such is the case of a Palisade GPS receiver which utilizes the event
>timestamping feature of a GPS receiver to provide as good (if not better)
>source as the PPS event capture on the computer side. Another clock
>mentioned was the ACTS (modem) reference.
>
>
Mine is a GPS18lvc with a PPS hookup usign the PPS built into the NMEA
driver.
>
>
>>Check out pogo.udel.edu, a Solaris connected to a Spectracom GPS and
>>PPS. I checked several days in the archive from October 2004 to now and
>>found no clockhops
>>
>>
>
>As far as I can see, pogo has no remote network ntp sources.
>Try adding several good network sources (as least 4), so they can
>contend with local clocks in finding the untruthful Byzantine general.
>
>Perhaps the problem is in a possible network driver asymmetry
>(there is no fudge1 offset option for network sources)
>giving all network sources the same systematic offset,
>making the fewer local refclocks lose the vote.
>
>
>
That makes sense to me. My system won't hop to stratum 2 network source
but will to stratum 1's (even if they are long RTT and jittery).
John
>
>
More information about the hackers
mailing list