Re: [gpsd-dev] PPS over USB

[Ed W]

On 08/05/2012 08:15, Eric S. Raymond wrote:
> Hal Murray<address@hidden>:
> > > One of the explicit goals of the Cosmic Backround Bufferbloat Detector is to
> > > sanity-check NTP.
> > If you see anything that seems unreasonable, please let me know.  ntpd has a
> > huge collection of heuristics (hacks?) that try to filter out all sorts of
> > quirks/glitches.  I'm sure there are some cases that will get past the
> > current filters.
> Oh trust me, if we find such phenomenae we are *not* going to be merely
> whispering about them - you and every other Internet infrastructure geek
> we know will get the A-OOGA A-OOGA red alert muy pronto. :-)

I would have expected the kinds of problems you are describing above to 
be more easily studied/tested using models in a lab?  Just turn on the 
various traffic shaping options and you can make your connection appear 
to be dropping packets/arbitrarily slow/turn up congestion.  I have a 
script that we use to pretend to be on an Iridium satellite dialup, ie 
2.4Kbits, ie 300 bytes per second - for sure we test everything over a 
real connection, but it sure is helpful to be able to stress things in 
certain ways

I don't really follow ntp development, but what I really like about 
chrony is every time you "have an idea" Miroslav agrees or disagrees 
with a whole bunch of investigative action and usually a graph or two.  
He really loves to go to the field and show the results!  He is also the 
chap who published some interesting numbers comparing the accuracy and 
offsets of a large number of machines in pool.ntp.org

Good luck

Ed W