[ntp:questions] Second attempt at GPS-18 based NTP server

David Lord snews at lordynet.org
Thu Feb 2 15:38:21 UTC 2012


David J Taylor wrote:
>> Hello All,
>>
>> I've (temporarily) given up on the idea of using Linux for the OS, and 
>> moved to FreeBSD, with the same GPS hardware, using some of the 
>> information on this web page:
>>
>> http://www.satsignal.eu/ntp/FreeBSD-GPS-PPS.htm
>>
>> ntpq -p now says:
>>
>>      remote           refid      st t when poll reach   delay   offset 
>> jitter
>> ============================================================================== 
>>
>> *dns1.rmplc.co.u 195.66.241.3     2 u   25   64    1    6.889    1.226 
>> 0.615
>>  91.220.127.250  130.88.200.6     3 u   24   64    1    5.600    4.940 
>> 0.274
>>  GPS_NMEA(1)     .PPS.            0 l   39   64    1    0.000    0.826 
>> 0.001
>>
>> So, I *think* that everything is okay, and it is using the PPS signal. 
>> There is one slight worry, which is what I see in /var/log/messages:
>>
>> Feb  2 09:50:16 tock ntpd[1573]: refclock_nmea: time_pps_kcbind 
>> failed: Operation not supported
>>
>> Please note, I have not done any recompiling of the kernel at this 
>> stage - just using the GENERIC kernel, because after reading the 
>> comments from Per Hederland in the System software customisation 
>> section of the above web page, it seemed unnecessary. Comments?
>>
>> Thanks for your help!
>>
>> Paul.
> 
> Paul,
> 
> I know little about FreeBSD, so I don't know what the error message 
> means. I would wait a little while, and if the "reach" for the 
> GPS_NMEA(1) reaches 377, I reckon you're OK.  On the Intel Atom system 
> here, it shows offset: 0.002 and jitter: 0.004 (these are in milliseconds).
> 
> I would have expected the poll for the GPS/PPS driver to be 16, and not 
> 64.  Here's an extract of what I have in the ntp.conf:
> 
> ________________________________
> server  127.127.20.1    mode 0  minpoll 4 maxpoll 4  prefer
> fudge   127.127.20.1    flag1 1 flag3 1 refid PPS
> 
> server    0.uk.pool.ntp.org    iburst minpoll 10 maxpoll 10
> server    1.uk.pool.ntp.org    iburst minpoll 10 maxpoll 10
> server    2.uk.pool.ntp.org    iburst minpoll 10 maxpoll 10
> ________________________________
> 
> 
> Forcing the Internet servers to be 1024s polling always.  I don't have 
> any note that I needed to recompile the kernel, so perhaps it was an 
> install option.
> 
> If you find any errors on the Web page, or there is something you would 
> like explained better, or to add, please tell me.  I'm glad you found it 
> helpful - well, some of it!


On NetBSD-5 with ntpd 4.2.6p5 the default is 64s.

I've just added the "minpoll 4 maxpoll 4" back to my ntp.conf
as it could reduce the 30 usec offset blips I get when logs are
rotated.

On my two pool servers I have named internet servers with
"minpoll 8 maxpoll 10 iburst" which gives reasonably fast
recovery from internet outages.

Temperature changes  here mean that on pcs that don't have
a PPS source "minpoll 10" just doesn't work and leads to
ntpd stepping the clock.


David



More information about the questions mailing list