[ntp:questions] Do logs indicate a config problem?

David Lord snews at lordynet.org
Fri Jul 16 22:02:53 UTC 2010


David Lord wrote:

> 16 July
> 09:56:27 ntpd exiting
> 09:56:39 clock PPS(2) event clk_noreply
> 10:01:01 synch to GPS_NMEA(2), stratum 0
> 10:01:01 kernel time sync status change 0x2101
> 10:02:02 kernel time sync status change 0x2107
> 10:02:17 synchronized to PPS(2), stratum 0
> 
> and "ntpq -p" gives PPS(2) offset 0.000ms and
> other servers on lan < 1.4ms.

OOPS

I had the parallel pps connected to where the watch xtal
osc had been disconnected so above is with normal serial
dcd pps from the gps.

I've now connected pps output from gps to the parallel
port with one of local servers as prefer.

oPPS(0)  .PPSb. 1 l .... 377 0.000 -0.010 0.011
.....
+server  .MSFa. 1 u .... 377 0.654  0.505 0.588

Using the gps serial output gives a known problem
where offset is too far out from the pps so requires
selection of a fudge time for pps to kick in. If
lucky pps is eventually selected.

oPPS(0)      .PPSb. 0 l .... 377 0.000 27.190 10.338
+GPS_NMEA(2) .GPSb. 0 l .... 377 0.000 42.004 10.348
.....
+server      .MSFa. 1 u .... 377 0.843 71.197 19.625

Another 20 minutes and pps offset has only fallen
to -3.3ms so startup and convergence can be a
lot slower with larger initial offset than using
a server on lan but stratum 0 is obtained with
offset of a few us.

Before someone chimes in, the fix does not seem
to be available in the ntpd source (NetBSD-5
STABLE has 4.2.4p6-o)


David




More information about the questions mailing list