[time-nuts] Is it sensible to update every few seconds from NTP server?

[Christopher Brown]

There are 2 different things here.


"Setting" the time based on a single query.

"Disciplining" the local clock


Many of the built in NTP clients just "Set" the time every X

Setting one of these to "SET" the local clock every X seconds is a less
than good thing.


If you timing needs are loose, let the client _set_ the time once an
hour or day.


If you need tight timing, install a full NTP setup.


Normally this means...


Host starts up

Host performs a _set_ to get the time within a few tens of ms

Host then fires up a proper ntp server, with a list of remote service.
This talks to all of the provided servers, figures the local osc offset,
compensates and keeps everything in-line.

This is a much better (and more stable) setup than hard setting the
clock every 4 seconds

On 11/7/12 12:41 PM, David Kirkby wrote:
> Someone at my radio club uses some mode of operation where accurate
> time is required. He said the standard Windoze clock does not keep
> sufficiently accurate, so he has software which updates from an NTP
> server every 4 seconds or so. It's not exactly a denial of service
> (DOS) attack, but seems almost close to it in NTP terms to me. I can't
> really believe updating every few seconds is sensible myself, but he
> assures me it works very well. (I'm rather hoping it does not use a
> stratum 1 server!)
> 
> I'm sure someone will say if you want accurate time on a PC, to use
> some combination of GPS, rubidium or OCXO with a 1 pps pulse and a
> serial port on a FreeBSD or similar computer. But that's probably not
> practical if your software only works on Windoze.
> 
> Any comments?
> 
> Dave, G8WRB.
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>