[ntp:questions] Thunderbolt at NTP ref clock.
unruh at invalid.ca
Tue Jul 30 19:35:15 UTC 2013
On 2013-07-30, David Lord <snews at lordynet.org> wrote:
> Dave Baxter wrote:
>> Hi again.
>> In reference to my other thread.
>> "THunderbolt monitor/control on Win7 ?"
>> Well, as that device seems happy in it's potential new home.
>> I've been poking about the interweb looking for info as to getting the
>> TB used as a reference clock for NTP.
>> I found this page:-
>> That on the face of it, seems to tell me what I need.
>> So, poking arround (no changes yet) my ntp.conf file, I "think" I need
>> # Trimble Thunderbolt on /dev/gps1 (first serial port)
>> server 127.127.29.1 mode 2 minpoll 4 maxpoll 4 prefer
>> fudge 127.127.29.1 time1 0.020
>> fudge 127.127.29.1 flag2 1 refid GPS
>> server 0.uk.pool.ntp.org iburst minpoll 5 maxpoll 12
>> server 1.uk.pool.ntp.org iburst minpoll 5 maxpoll 12
>> server 2.uk.pool.ntp.org iburst minpoll 5 maxpoll 12
>> Would this be suitable for general purpose local time sourcing? With
>> fallback to the pool if the TB burps, dies or is otherwise compromised.
> I'd add a few more public servers, I've been using five.
Why not 47?
His are backups. 3 is fine.
>> Could all the fudge factors go on one line?
> Shouldn't be a problem.
> fudge 127.127.29.1 time1 0.020 flag2 1 refid GPS
That looks like a huge time1 fudge. Are you sure that the pool servers
do not actually give better time than does the Trimble?
>> (Sorry to ask silly questions, but I don't know, so have to ask.)
>> If so, in what (if any) order?
>> The server currently run's FreeBSD 8.0 Uptime over 772 days now,
>> without a hitch! Nice... (I do plan to configure another box, running
>> 9.x at some point to replace this one.)
>> ntpq -crv reports (among other things)...
>> version="ntpd 4.2.4p5-a (1)"
> I'm using 4.2.6p5 but that is fairly old (Dec 2011 maybe). One
> of the 4.2.7pnnn may be a better choice although I tried both
> 4.2.4p354 and p359 on NetBSD and had problems using the stom
More information about the questions