[ntp:questions] "Large" offsets

Miguel Gonçalves mail at miguelgoncalves.com
Thu Aug 4 08:26:53 UTC 2011


Hi there!

I have a Garmin 18 LVC connected to a FreeBSD 7.4 machine and I am
getting "high" offsets. Here are the stats of yesterday


NTP Loopfile Analysis
---------------------

 0 us:    6 samples ( 0.12%) ( 0.12%)
 1 us:   21 samples ( 0.41%) ( 0.53%)
 2 us:   33 samples ( 0.65%) ( 1.18%)
 3 us:   58 samples ( 1.15%) ( 2.33%)
 4 us:  109 samples ( 2.15%) ( 4.48%)
 5 us:  270 samples ( 5.33%) ( 9.81%)
 6 us:  490 samples ( 9.68%) (19.49%)
 7 us:  650 samples (12.84%) (32.33%)
 8 us:  532 samples (10.51%) (42.84%)
 9 us:  531 samples (10.49%) (53.33%)
10 us:  677 samples (13.37%) (66.70%)
11 us:  793 samples (15.66%) (82.37%)
12 us:  543 samples (10.73%) (93.10%)
13 us:  226 samples ( 4.46%) (97.56%)
14 us:  107 samples ( 2.11%) (99.67%)
15 us:   15 samples ( 0.30%) (99.97%)
16 us:    1 samples ( 0.02%) (99.99%)

5062 / 5062 samples evaluated

The GPS is on the roof and has a 10 meter long cable to connect to the
server. Can this be the cause? How to improve?

By the way, I have another machine with a Sure GPS board and I get this:


NTP Loopfile Analysis
---------------------

 0 us: 1977 samples (36.61%) ( 36.61%)
 1 us: 1739 samples (32.20%) ( 68.81%)
 2 us:  961 samples (17.80%) ( 86.61%)
 3 us:  455 samples ( 8.43%) ( 95.03%)
 4 us:  181 samples ( 3.35%) ( 98.39%)
 5 us:   42 samples ( 0.78%) ( 99.16%)
 6 us:   23 samples ( 0.42%) ( 99.59%)
 7 us:   17 samples ( 0.31%) ( 99.90%)
 8 us:    5 samples ( 0.09%) ( 99.99%)

5400 / 5400 samples evaluated

I mean 15 us are not bad but less than 10 is better (and most of the
time 0 or 1 us).

Thanks in advance for any help.

Regards,
Miguel



More information about the questions mailing list