Why is root dispersion so high ?

[Horia Muntean]

Hi,

I configured time3.google.com as a time source for a chrony ntp server. 

What I observed from the measurements done during operation is that the 'Root Dispersion' values reported by the google server are quite high for a stratum-1 server. 

========================================================================================================================

   Date (UTC) Time     IP Address   L St 123 567 ABCD  LP RP Score    Offset  Peer del. Peer disp.  Root del. Root disp.

========================================================================================================================

2017-09-04 08:42:44 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.736e-03  4.889e-02  3.660e-06  0.000e+00  9.003e-04

2017-09-04 08:42:48 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.671e-03  4.764e-02  3.541e-06  0.000e+00  2.060e-03

2017-09-04 08:42:52 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.358e-03  4.405e-02  3.330e-06  0.000e+00  1.297e-03

2017-09-04 08:42:56 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.006e-03  4.884e-02  3.281e-06  0.000e+00  2.396e-03

2017-09-04 08:43:00 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.656e-03  4.747e-02  3.186e-06  0.000e+00  8.087e-04

2017-09-04 08:43:04 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.691e-03  4.356e-02  3.031e-06  0.000e+00  2.426e-03

2017-09-04 08:43:09 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.557e-03  4.767e-02  2.950e-06  0.000e+00  8.545e-04

2017-09-04 08:43:13 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.227e-03  4.586e-02  2.882e-06  0.000e+00  7.172e-04

2017-09-04 08:43:17 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.928e-03  4.842e-02  2.881e-06  0.000e+00  7.782e-04

2017-09-04 08:43:21 216.239.35.8    N  1 111 111 1111   2  2 1.00 -2.017e-03  4.305e-02  2.761e-06  0.000e+00  8.545e-04

2017-09-04 08:43:25 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.560e-03  4.491e-02  2.702e-06  0.000e+00  3.464e-03

2017-09-04 08:43:29 216.239.35.8    N  1 111 111 1111   2  2 1.00 -8.420e-04  4.343e-02  2.632e-06  0.000e+00  9.918e-04

2017-09-04 08:43:33 216.239.35.8    N  1 111 111 1111   2  2 1.00 -4.080e-04  4.600e-02  2.634e-06  0.000e+00  8.392e-04

2017-09-04 08:43:37 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.675e-03  4.368e-02  2.585e-06  0.000e+00  9.613e-04

2017-09-04 08:43:41 216.239.35.8    N  1 111 111 1111   2  2 1.00 -8.720e-04  4.529e-02  2.553e-06  0.000e+00  8.545e-04

2017-09-04 08:43:46 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.956e-03  4.842e-02  2.569e-06  0.000e+00  6.866e-04

2017-09-04 08:43:50 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.614e-03  4.920e-02  2.581e-06  0.000e+00  3.922e-03

2017-09-04 08:43:54 216.239.35.8    N  1 111 111 1111   2  2 1.00 -3.631e-03  4.907e-02  2.575e-06  0.000e+00  2.625e-03

2017-09-04 08:43:58 216.239.35.8    N  1 111 111 1111   2  2 1.00 -1.990e-03  4.439e-02  2.510e-06  0.000e+00  9.155e-04

Never mind the high offset recorded as another local server was chosen to be the current time source, my question refers only to the 'root dispersion' reporter by the google server.

Even if one is on the same LAN with such a server and suppose the offset would be zero and 'peer delay' would be much lower, the accuracy to UTC true time on the client would be ruined by a 'root dispersion' value that high.

Can someone explain why the values are so high ? 

BR

[Chris Schrier]

Greetings,

Thank you for bringing this to our attention. We have started discussing this and agree that something seems off. We hope to have a better answer in a few days. In the meantime, would you mind filing an entry for this in our issue tracker? 

https://issuetracker.google.com/issues/new?component=180620

I can file an issue if you're unable to for some reason.

Thanks again.

--
You received this message because you are subscribed to the Google Groups "public-ntp-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to public-ntp-disc...@googlegroups.com.
To post to this group, send email to public-nt...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/public-ntp-discuss/147e524d-7a98-4d16-b287-4c212e97a752%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Horia Muntean]

Here it is: https://issuetracker.google.com/issues/65392996

Thanks

To unsubscribe from this group and stop receiving emails from it, send an email to public-ntp-discuss+unsub...@googlegroups.com.