Re: Time offset / Jitter under FreeBSD




"Dag-Erling Smørgrav" <des@xxxxxx> wrote in message
news:86fxykdm4z.fsf@xxxxxxxxxxxxx
"Bob" <bobsjunkmail@xxxxxxxxxxxxx> writes:
I installed FreeBSD on a drive, and booted my windows machine from it.
The
FreeBSD port of ntpd shows a offsets in the 20-60 ms range. Issues with
jitter, also. Remember, the same machine running windows is generally sub
millisecond - http://www.pool.ntp.org/scores/74.181.17.243/log offset.

What does 'sysctl kern.timecounter' say?

DES
--
Dag-Erling Smørgrav - des@xxxxxx


What we have here is BSD noob coupled with lack of patience. That, coupled
with the fact that the Windows version seems to have a faster slew rate. I
left it alone long enough, and all is well. The Windows version seems to
figure out the computer clock offset alot faster. The "default" location of
the ntp.drift file is not in /etc. It is of course in /var/db, so all the
looking and fiddling i was doing to the one in /etc was for nothing. I did
build, and am running 4.2.4p4@xxxxxxxx

It's behaving well, except it doesn't like to read my NTP-100 (10.33.90.50),
while the Meinberg windows version (10.33.90.10) does just fine.

BTW, this machine has no disk. It is running on an off the shelf 2 gig USB
memory stick. 512 meg ram, no swap slice. It appears not to access the stick
very often - maybe once every few minutes - but I'm not sure if they are
reads or writes. Anyone here that can tell me if there is a way to see how
often disk writes are occurring? I'd like to see how fast I'm burning up the
flash stick. It's a full minimal install + man pages. The current version of
ntpd was built on this machine.

Thanks for the help.

n2qew# ntpq -p
remote refid st t when poll reach delay offset
jitter
==============================================================================
-ns.unc.edu 129.6.15.28 2 u 6 64 367 46.649 5.865
1.138
+64.236.96.53 .ACTS. 1 u 2 64 317 39.811 2.366
0.269
+68.216.79.113 .ACTS. 1 u 21 64 377 21.753 3.099
0.283
*10.33.90.10 10.33.90.50 2 u 4 16 377 0.211 -0.003
0.025
10.33.90.50 .INIT. 16 u - 64 0 0.000 0.000
0.000
n2qew#

n2qew# ntpq -p 10.33.90.10
remote refid st t when poll reach delay offset
jitter
==============================================================================
-ns.unc.edu 129.6.15.28 2 u 46 64 357 46.198 5.775
0.543
+64.236.96.53 .ACTS. 1 u 22 64 357 38.828 2.178
0.593
+68.216.79.113 .ACTS. 1 u 86 64 376 21.989 3.213
0.719
*10.33.90.50 .GPS. 1 u 7 16 377 0.428 0.057
0.005
n2qew#

ntpdc> kerni
pll offset: -7.889e-06 s
pll frequency: 19.482 ppm
maximum error: 0.021753 s
estimated error: 8e-06 s
status: 2001 pll nano
pll time constant: 4
precision: 1e-09 s
frequency tolerance: 496 ppm
ntpdc>

n2qew# sysctl kern.timecounter
kern.timecounter.stepwarnings: 0
kern.timecounter.nbinuptime: 12748095
kern.timecounter.nnanouptime: 0
kern.timecounter.nmicrouptime: 2041
kern.timecounter.nbintime: 113343
kern.timecounter.nnanotime: 56018
kern.timecounter.nmicrotime: 19303
kern.timecounter.ngetbinuptime: 842
kern.timecounter.ngetnanouptime: 1471
kern.timecounter.ngetmicrouptime: 909935
kern.timecounter.ngetbintime: 0
kern.timecounter.ngetnanotime: 0
kern.timecounter.ngetmicrotime: 6558219
kern.timecounter.nsetclock: 4
kern.timecounter.hardware: ACPI-fast
kern.timecounter.choice: TSC(800) ACPI-fast(1000) i8254(0) dummy(-1000000)
kern.timecounter.tick: 1
n2qew#




.



Relevant Pages

  • Re: refclock on Windows graphs
    ... last 72 hours the frequncy varied between about -0.6ppm to +0.4ppm from the mean, and on the Windows XP system from about -1.4ppm to +1.1ppm from the mean. ... So about 1ppm variation as opposed to 2.5ppm, about 40% of the frequency variation. ... As far as I can tell, the NTP offset is more a function of the control loop parameters than of the actual hardware (with ~150us offsets and yet only 2.5us jitter). ...
    (comp.protocols.time.ntp)
  • Re: Tighter regulation?
    ... If measurement errors are variable, ... I am not sure what jitter will do, but what good is low jitter at the ... Increased jitter means that your offset measurements span a wider range ... Windows load related errors, ...
    (comp.protocols.time.ntp)
  • Re: ALC883 recording troubles...
    ... On Wed, Jun 11, 2008 at 8:00 PM, Daniel J Blueman ... I'm experiencing DC offset with the microphone on 2.6.24 (Ubuntu 8.04 ... Would it help to install windows, ...
    (Linux-Kernel)
  • Re: Tighter regulation?
    ... above 256 seconds and that allows excessive wandering of my clock. ... latencies, and the difficulties of interpolating clock ticks, and writing ... The precision figure, for Windows, is almost ... It is an offset from the best time source I have available. ...
    (comp.protocols.time.ntp)
  • Re: Win7: ntpd adjusting time backwards
    ... Loopstats collection has been on since the beginning. ... Windows Vista or Windows-7. ... "using Windows clock directly" in the NTP events. ... and initially the offset stayed under 10 ms, but it seems that as soon ...
    (comp.protocols.time.ntp)