Test farm news

Gary E. Miller gem at rellim.com
Mon Jun 13 22:32:26 UTC 2016

Yo Eric!

On Mon, 13 Jun 2016 17:59:16 -0400
"Eric S. Raymond" <esr at thyrsus.com> wrote:

> Another donation arrived today - an Uputronics HAT and ModMyPi case
> made for it, with an exit hole for the SMA jack. (Thank you, Anthony
> Stirk!) I have ordered another Pi 3 to go with it.

Cool!  then you can get some good A/B testing in.

> The main reason I haven't started running tests already is that I
> don't yet understand what metrics to watch, nor what the good
> profiling and data-reduction tools are.

Local Clock Offset is the big one, the whole point of ntpd is to get
the local clock slaved to the PPS.

Everything else is just working towards that goal.  There will be more things
worth measuring/displaying, but until the startup transients are
gone not worth looking at, yet.

Except, I am finding over three machines, that every few days a PPS will 
pop out by one second, then mess up things for hours.  I got work to
do in gpsd...

> Would somebody please educate me about this?  Pointers to things I
> could read would be helpful.

Check these out:

Those are graphs and calculations from the ntpd stats files.

Things like NMEA offsets, startup wobbling, the quality of USB versus
serial GPS, and much more is in there.

Those are based on Dan Drown's work from here:

There are other similar projects.

Don't worry about what the graphs mean, you need a weeks data graphed to
see the full effect.  You have been running ntpd stats logging, right?

Then going forward the restart effects will be jarring to you.

Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
	gem at rellim.com  Tel:+1 541 382 8588
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 473 bytes
Desc: OpenPGP digital signature
URL: <http://lists.ntpsec.org/pipermail/devel/attachments/20160613/a462e808/attachment.bin>

More information about the devel mailing list