On Sat, Jan 28, 2012 at 22:02, Kasper Pedersen <[email protected]> wrote:
>
> Can anyone give me a hint to understanding this difference between root
> dispersion, as reported by the production version, and the dev version,
> of ntpd?
>
> My test machine has a very stable oscillator, and a good PPS source.
> configuration is below:  a PPS source, gpsd though SHM, and a local server.
>
> When running the production (4.2.6p5) ntpd, I see root dispersion settle
> to a handful of microseconds. It increments at the dispersion (15us/s),
> as documented, until pulled back down on the next update.
>
> When running the DEV ntpd, root dispersion is pulled down to 1000us, and
> increments at 15us/s. No matter what, rootdisp does not go below 1ms.
>
> [root@n1g ntpq]# ./ntpq -crv
> associd=0 status=011d leap_none, sync_pps, 1 event, kern,
> version="ntpd [email protected] Sat Jan 28 11:01:39 UTC 2012 (1)",
> processor="x86_64", system="Linux/3.1.8-2.fc16.x86_64", leap=00,
> stratum=1, precision=-23, rootdelay=0.000, rootdisp=1.000, refid=PPS,
> reftime=d2ceed35.e7967bf1  Sat, Jan 28 2012 21:55:01.904,
> clock=d2ceed38.1b41ac47  Sat, Jan 28 2012 21:55:04.106, peer=60158, tc=3,
> mintc=3, offset=-0.000, frequency=38.361, sys_jitter=0.001,
> clk_jitter=0.000, clk_wander=0.000, tai=34, leapsec=201207010000,
> expire=201212280000
>
>
> Is this by design, or an oops? I stumbled upon it and though 'this might
> not be right'.

This is by design.

ChangeLog has under 4.2.7p237:
* Floor calculation of sys_rootdisp at sys_mindisp in clock_update (from
  Dave Mills).

sys_mindisp is controlled by "tos mindist" and defaults to 1 millisecond:

[1] http://www.eecis.udel.edu/~mills/ntp/html/miscopt.html#tos

Cheers,
Dave Hart
_______________________________________________
questions mailing list
[email protected]
http://lists.ntp.org/listinfo/questions

Reply via email to