[ntp:questions] Ignore one server except in extreme conditions?

David Lord snews at lordynet.org
Wed Nov 16 11:20:49 UTC 2011


A C wrote:
> Is there a way to configure ntpd to ignore a particular clock unless 
> there is no other choice?  What I'm thinking is to ignore the GPS 
> receiver NMEA data and use only the PPS plus Internet servers for most 
> of the synchronization.  But if the Internet servers go down, accept the 
> GPS NMEA data for seconds numbering and let the tick be controlled by PPS.

Doesn't ntp do that already?

If you have nmea + pps the pps will be used for sync.
If your gps + pps is down your internet servers will be used for sync.


  Wed Nov  2 00:00:00 GMT 2011
       remote           refid      st t when poll reach   delay   offset 
  jitter
 
==============================================================================
  *GPS_NMEA(2)     .GPSb.           0 l   29   64  377    0.000  -52.888 
  11.213
  oPPS(2)          .PPSb.           0 l   28   64  377    0.000   -0.001 
   0.004
   p4x2400b.home.l 192.168.59.61    2 u    6   64  377    1.157   -3.917 
   0.908
   p4x2400c.home.l 192.168.59.61    2 u   19   64  377    0.729    2.298 
   0.619
   pd6000e1.home.l 192.168.59.61    2 u   11   64  377    0.544    1.767 
   0.551
   pd6000e2.home.l 192.168.59.61    2 u   54   64  377    0.494    0.983 
   0.535
  +ntp1.lordynet.o 81.187.61.74     2 u    7  128  377    0.854    2.089 
   0.520
  +ntp0.lordynet.o .MSFa.           1 u    1  128  377    0.934    1.548 
   0.572

  offset:               -0.000001 s
  frequency:            -35.833 ppm
  poll adjust:          30
  watchdog timer:       28 s
  ntp_gettime() returns code 0 (OK)
    time d25b0680.8615a8ac  Wed, Nov  2 2011  0:00:00.523, (.523768144),
    maximum error 14934 us, estimated error 4 us, TAI offset 34
  ntp_adjtime() returns code 0 (OK)
    modes 0x0 (),
    offset 1.094 us, frequency -35.833 ppm, interval 256 s,
    maximum error 14934 us, estimated error 4 us,
    status 0x2107 (PLL,PPSFREQ,PPSTIME,PPSSIGNAL,NANO),
    time constant 6, precision 0.001 us, tolerance 496 ppm,
    pps frequency -35.833 ppm, stability 0.020 ppm, jitter 2.688 us,
    intervals 17356, jitter exceeded 15439, stability exceeded 0, errors 19.


David

> 
> I guess I'm asking for something opposite to the "prefer" flag, a "not 
> unless you have to" flag.  The GPS does get selected among the various 
> clock sources but it's the worst of the bunch with respect to jitter and 
> offset.  Those two wander all over while the rest of the Internet 
> servers and the PPS are stable.  It is almost never rejected as an 
> outlier or false ticker even if it's way out of line with the other 
> sources.



More information about the questions mailing list