[ntp:questions] ntpdate.c unsafe buffer write

Unruh unruh-spam at physics.ubc.ca
Mon Feb 11 18:57:22 UTC 2008

jason at extremeoverclocking.com (Jason Rabel) writes:

>I've tried to keep quiet and bite my tongue at this whole ntp vs chrony
>thing... But something has been nagging me in the back of my head that i
>juat have to know the answer to...

>How are you measuring your results? From what I've skimmed over you are
>simply using each program's own generated statistics... Wouldn't a more
>correct way be to use an external (and calibrated) device to measure /
>compare to ensure the results are actually valid? Otherwise you are in
>essence comparing apples to oranges...

Well, no. Yes, I do use each programs statistics, but they are the raw
statistics of the offset (t1+t4-t2-t3)/2 and the round trip (t2-t1+t4-t3)
from the ntp packets. There is no processing that has gone on in eitehr
system in reporting those numbers. In fact in ntp I hacked the
report_peerstats function to report the raw numbers, not the ones that have
gone through the clock filter.
If the statistics of those are the same for the two, one cannot say
anything. If one is worse than the other however ( which is what I find for
ntp on the internal network) then the one that is worse is a worse clock. 

Yes, ideally one would also have an independent external time source which
one uses to see how well either system disciplines the clock to true utc,
and I am getting one to look at the statistics for another system I have,
which is connected to my server ( A GPS PPS driven computer running ntp and
getting a variance of about 2-3usec) via an ADSL line from home ( 20msec
round trip typical and about .5ms variance for both programs for the "raw"
offset distribution. Well, not quite raw, but with roundtrips >2minimum
removed from the statistics). The two programs are very similar, so I
cannot say which disciplines the clock better. Eg, ntp might discipline it
to within 5usec of UTC while chrony only does it to within 100usec and I
could not tell by looking at the variance, since I do not believe that the
variance is accurate to 10%.

So, no, I am comparing apples to apples ( the offsets as determined from
the ntp packet exchange mechanism which both use and both report). 

More information about the questions mailing list