Test farm news

Gary E. Miller gem at rellim.com
Tue Jun 14 18:05:48 UTC 2016

Yo Eric!

On Tue, 14 Jun 2016 06:32:36 -0400
"Eric S. Raymond" <esr at thyrsus.com> wrote:

> Gary E. Miller <gem at rellim.com>:
> > > I can easily shut down and restart all ntpds - in fact I'll have
> > > to in order to get the new configs read.  
> > 
> > You need to do more than shutdown ntpd to get back to pre-ntp
> > initial conditions.  For a real good test you need to set the
> > system clock off by maybe a minute, or an hour ,or maybe back to
> > 1970.  Remove the drift file.  Remove the battery from you GPS.
> > And probably restart Linux to clera out any save drift.  
> /me imagines hitting something with a cartoon mallet.
> That would be the most extreme test, but for that very reason I doubt
> it's that interesting - how often are you going to get hit with all
> of this at once? Still...

Which is exactly why ntpd behaves so badly now on startup.  No one
involved with ntpd ever took this case seriously.  This is almost
exactly the case you will get every time you start a laptop or RasPi
after moving it.  For comparision, chronyd goes to great lengths to
handle this case well, and it shows.

> # This does everyting but the battery removal.

Do not underestimate how badly cold start can make the GPS perform.

> ssh -c "sudo '(date -s 0; rm /var/lib/ntp/ntp.drift; reboot)'" $1
> I shall call it "mallet".

I also neglected to mention letting your laptop/raspi cool down.  I notie
that the clock switcngs wildly when warming up and that also confuses
the PLL to do perverse things.

And none of this will do you any good until you wait another 7 days.  If
you had turned on logging last week you would have the data to look at

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/20160614/38a91e13/attachment.bin>

More information about the devel mailing list