[ntp:questions] Poll interval stuck
Brian.Inglis at SystematicSw.ab.ca
Mon Mar 28 16:19:23 UTC 2016
On 2016-03-27 15:49, Geoff Down wrote:
> On Mon, Jan 18, 2016, at 04:47 PM, Brian Inglis wrote:
>> On 2016-01-18 03:07, Geoff Down wrote:
>>> On Mon, Jan 18, 2016, at 01:54 AM, Brian Inglis wrote:
>>>>>> On Fri, Jan 15, 2016 at 7:54 AM, Geoff Down <geoffdown at fastmail.net>
>>>>>>> (using the default config from Macports, so no minpoll set).
>>>> On 2016-01-16 17:39, Geoff Down wrote:
>>>>> Thanks both, its a PowerMac PPC (so not recent....)
>>>>> remote refid st t when poll reach delay offset
>>>>> *ussjc2-ntp-002. .GPSs. 1 u 63 64 377 153.056 1.119
>>>> Instead of the default config e.g.
>>>> server time.apple.com
>>>> which appears to have selected a server on the other side of a continent
>>>> ocean from you (153ms delay!) - Cupertino, California, USA
>>>> you might want to change that to e.g.
>>>> pool time.apple.com iburst preempt
>>>> pool pool.ntp.org iburst preempt
>>>> or even
>>>> pool CC.pool.ntp.org iburst preempt
>>>> where CC is your two character ISO country code e.g. us...
>>>> in the hopes that ntpd can pick some better server(s) closer to you.
> To revisit this:
> poll interval is still stuck at 64s even though I have the following
> conf file (which predates the startup of ntpd):
> "# Limit network machines to time queries only
> restrict default kod nomodify notrap nopeer noquery
> restrict -6 default kod nomodify notrap nopeer noquery
> # localhost is unrestricted
> restrict 127.0.0.1
> restrict -6 ::1
> disable monitor
> server time.euro.apple.com iburst minpoll 10"
Hopefully the line above reflects your actual conf file?
All the server options have to be on the same line.
> remote refid st t when poll reach delay offset jitter
> *uklon5-ntp-001. .GPSs. 1 u 2 64 377 22.285 0.874 0.108
The minpoll and maxpoll mainly act to limit the network load by increasing the
burst reply timeout and unreachable server retry interval.
Poll 10 is only reached if your frequency drift settles to a very stable level
over a long period of time, or is disciplined externally, such as by an atomic
clock source, which is why 10 is the default for *maxpoll*.
To get poll to 10, your time must be initially sampled much more frequently to
calculate your frequency drift and discipline your clock.
The poll process uses the quality of the time samples it gets to control the
poll interval, and over a period of days may settle down to poll 10, with a
stable system clock and good network sources.
In your case, with system clock and single network source of unknown quality,
it will probably stick at the built in default minpoll of 6.
Suggest removing your minpoll, add a few more network sources such as:
server ntp1.npl.co.uk iburst
server ntp2.npl.co.uk iburst
pool uk.pool.ntp.org iburst
and any other close public network sources you can find, such as your ISP,
local universities, and from lists .
Then monitor your system frequency drift and offset hourly over a few days
$ ntpq -c rv
associd=0 status=041d leap_none, sync_uhf_radio, 1 event, kern,
version="ntpd 4.2.6p5 at 1.2349-o Mon Mar 21 18:34:20 UTC 2016 (1)",
processor="armv7l", system="Linux/4.1.19-v7+", leap=00, stratum=1,
precision=-20, rootdelay=0.000, rootdisp=0.474, refid=GPS,
reftime=daa3cbe1.e424d27b Mon, Mar 28 2016 16:27:29.891,
clock=daa3cbef.fad2ee9b Mon, Mar 28 2016 16:27:43.979, peer=42816, tc=4,
mintc=3, offset=-0.029, frequency=-5.305, sys_jitter=0.002,
^^^^^^ ^^^^^^ ^^^^^
clk_jitter=0.000, clk_wander=0.007, tai=36, leapsec=201507010000,
to see if your system offset, frequency drift, system and clock jitter,
and clock wander improve then stabilize.
 http://doc.ntp.org/current-stable/poll.html for more details
for UK servers
scroll down to GB and click on each server entry to find addresses
and if there are any access restrictions (most require only email
notification to the server admin).
Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada
More information about the questions