Date: Tue, 30 Jan 96 23:10 PST From: pete@pelican.com (Pete Carah) To: questions@freebsd.org Subject: Re: good NTP servers Message-ID: <m0thWgP-0000SMC@pelican.com> In-Reply-To: <199601261722.LAA12311@base486>
next in thread | previous in thread | raw e-mail | index | archive | help
In article <199601261722.LAA12311@base486> dave writes:
>> A nice list can be found at ftp.udel.edu somewhere in /pub/ntp -- it might
>> be in the xntpd distribution itself. Lots of nice stratum 2 servers to
>> sync to.
>I looked -- there's supposed to be a file "clock.txt" but it wasn't on
>ftp.udel.edu /pub/ntp/{.,doc}... Does anyone know where one can get
>a list of time servers who won't get upset if people use them to sync
>up their personal machines?
Strange - it's gone!! I also fetched the latest (3.4y) dist and it
isn't in there either. I have one fetched a while ago (Jan 95; it is
sort of out of date)... I can't find a more recent one on my other
sites, either.
The USNO's web site contains a pair of pages containing clock.txt (marked
as stolen from Dave Mills so it must still be somewhere on louie...); the
stratum 2 list is marked as updated 24 Jan 1996 so is pretty current.
see http://tycho.usno.navy.mil/clocks2.html (and clocks1.html if you want
stratum 1...)
You can use ns.clubnet.net or club.pelican.com but they are usually
stratum 3. The stratum 2 servers I tend to use are maui.cs.ucla.edu,
tick.cs.unlv.edu, tock.cs.unlv.edu, a couple at JPL and a couple in Oregon.
The *real* stratum 1 servers are tick.usno.navy.mil and tock.usno.navy.mil;
for a description of how they work, check http://www.usno.navy.mil/
and look at the "directorate of time".
(they are hppa machines with modified hardware and software clocks,
sync'd to one of the same h-maser sources that the gps timebases are
sync'd to.) (h-maser is more short-term stable than cesium; the real
definition of the second is cesium (I think) but those oscillators are
more complicated since the gas cell is used to phase-lock a crystal
with a fairly long loop time constant (minutes to hours).)
This list shows clepsydra.dec.com as stratum 2; it is sometimes stratum 1.
You can check out a host listed in the clock.txt file (or wherever) using
ntpq; for example, using one of the hosts given in clocks2.html:
ntpq> host tick.cs.unlv.edu
current host set to elmore.cs.unlv.edu
ntpq> peers
remote refid st t when poll reach delay offset disp
==============================================================================
-kks1.lbl.gov clock.llnl.gov 2 u 26 128 376 21.41 37.756 34.18
-cedr.lbl.gov clock.llnl.gov 2 u 31 256 377 27.24 -40.551 13.82
-quench.lbl.gov clock.llnl.gov 2 u 9 128 377 35.86 -29.701 10.56
.chico.rice.edu otc1.psu.edu 2 u 53 64 176 100.74 -25.787 443.89
.grainne.lbl.gov clock.llnl.gov 2 u 13 64 37 32.99 -40.009 946.32
.cheetah.llnl.go clock.llnl.gov 2 u 479 1024 176 52.67 -41.972 20.19
*time-A.timefreq .ACTS. 1 u 251 1024 377 61.78 -3.718 4.67
fuzz.sdsc.edu afarm-1.sdsc.ed 3 u 300 1024 377 108.81 -36.660 6.99
+GRAPEFRUIT.SRV. haven.umd.edu 2 u 442 1024 377 113.27 -2.656 7.25
.little-walter.C Sony.COM 2 u 1002 512 174 1.72 -3.500 135.60
LOCAL(0) LOCAL(0) 8 l 22 64 377 0.00 0.000 13.90
broadcast 0.0.0.0 16 u - 64 0 0.00 0.000 16000.0
-uncle-sam.llnl. clock.llnl.gov 2 u 289 512 376 45.33 -50.506 12.99
.samson.llnl.gov clock.llnl.gov 2 u 290 512 376 64.62 -53.588 13.15
-order.llnl.gov clock.llnl.gov 2 u 736 1024 376 40.05 -42.875 15.17
+204.99.130.2 ncar.ucar.edu 2 u 140 1024 377 73.79 -5.926 6.76
+cujo.Colorado.E ncar.ucar.edu 2 u 128 256 356 63.02 -10.671 7.63
.tornado.nwa.com otc1.psu.edu 2 u 823 1024 276 234.44 -26.810 24.44
.gandalf.sigmaso tick.usno.navy. 2 u 93 64 136 150.62 -40.367 15.21
boulder.Colorad clepsydra.dec.c 2 u 41 256 323 89.63 -21.356 2012.45
ntpq>
-----------------------------------------------------------------------------
(delay, offset, and disp are all in milliseconds); reach is a bitmask filled
from the right. When and poll are seconds. In the left column, + are sync
candidates and * is the current sync master.
Not all hosts in this peers list are public.
SGI's running IRIX 5.1 or 5.2 are *not* good candidates to sync to. I don't
know about 5.3; it was supposed to be better. 4.0.x are pretty good.
FreeBSD is pretty good if it has a good net connection; a compressing
modem with an asymmetric load like a newsfeed is *not* a good connection.
The net regions are no longer determined geographically; you have to
know your provider's connectivity (my two above are T1-connected to
mci in Los Angeles)...
-- Pete
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0thWgP-0000SMC>