[LinuxPPS] outlier

[Udo van den Heuvel]

Udo van den Heuvel wrote:
> # ntpq -pn
>      remote           refid      st t when poll reach   delay   offset
> jitter
> ==============================================================================
>  127.127.1.0     .LOCL.          10 l   19   64  377    0.000    0.000
>  0.001
> x127.127.20.0    .GPS.            0 l    6   16  377    0.000  -172.16
>  1.298
> +194.109.22.18   129.132.2.21     3 u  169  512  377   10.163   -1.852
>  7.934
> +193.67.79.202   .GPS.            1 u  108  512  377   32.090    4.631
>  7.402
> *193.79.237.14   .GPS.            1 u  173  512  377   13.634   -3.622
> 14.154
> 
> 
> Still my gps is an outlier.
> flag2 1 or 0 does not relaly matter.
> also, as I wrote, the proc assert/clear entries remain 0.
> I did reboot into 2.6.19 with the latest patch.
> No change. What could be wrong?
> 
> The PPS is 200ms so it is not exactly just an edge problem...

When I use a 200 ms compensation like this:

fudge 127.127.20.0 flag3 1 flag2 0 time1 0.200


I get:

# ntpq -pn
     remote           refid      st t when poll reach   delay   offset
jitter
==============================================================================
 127.127.1.0     .LOCL.          10 l   53   64  377    0.000    0.000
 0.001
*127.127.20.0    .GPS.            0 l    9   16  377    0.000   -7.342
 0.682
-194.109.22.18   129.132.2.21     3 u  372 1024  377    7.918  -30.701
 4.760
+193.67.79.202   .GPS.            1 u  371 1024  377   17.898  -33.376
 2.238
+193.79.237.14   .GPS.            1 u  236 1024  177   12.983  -36.769
12.073

So somewhat better but still with 30 ms offsets for rather accurate
remote clocks.