[ntp:questions] Tutorial for setting up Garmin 18 LVC on FreeBSD 8.0
unruh at wormhole.physics.ubc.ca
Tue Feb 23 00:58:51 UTC 2010
On 2010-02-23, ryandoyle <rd at ryandoyle.net> wrote:
> On Feb 23, 5:13?am, unruh <un... at wormhole.physics.ubc.ca> wrote:
>> On 2010-02-22, ryandoyle <r... at ryandoyle.net> wrote:
>> > For anyone interested in setting up a Garmin 18 LVC GPS receiver on
>> > FreeBSD 8, I wrote up a fairly detailed tutorial of my experiences
>> > here:http://blog.doylenet.net/?p=145
>> > A couple of months ago I thought it would be a neat idea to run a
>> > strat 1 time server at my work. I hadn't had any experience with NTP
>> > before, didn't know what local clock drivers were etc... and coming
>> > from a Linux background I had very little experience with FreeBSD.
>> I disagree with your comments re the accuracy on Linux. I have
>> consistantly gotten 2usec accuracy on Linux with ntpd, (and with chrony, Lichvar
>> gets sub usec accuracy with no change in the kernel).
> Admittedly, the wild inaccuracies that I was experiencing could have
> been due to my setup. I was running GPSd and using it as a local clock
> driver for NTP. At the time is seemed the "cleanest" way. Also I found
> that FreeBSD becomes stable at a much quicker rate compared to Linux.
That is a bit strange, since the algorithm used under linux is identical
to that under BSD. Was this on the same machine? If not, it might be that
one machine had larger temperature variations than the other.
> I never like patching the Linux kernel when package management is very
> tightly integrated into the distribution such as RHEL/CentOS which I
> was using and is why I didn't try other methods such as those found
> here: http://time.qnan.org/.
I agree. But I have no idea why kernel patching is needed.
adjtimex system calls seem more than sufficient.
More information about the questions