| Re: [chrony-users] Comparing timekeeping performance with chronyd to ntpd |
[ Thread Index |
Date Index
| More chrony.tuxfamily.org/chrony-users Archives
]
On Mon, Apr 29, 2019 at 05:27:12PM -0700, Watson Ladd wrote:
> I'm trying to compare the performance of ntpd and chronyd in our
> environment but am running into a snag. Our first thought was to look
> at the offset measurements taken by each, but they don't seem to be
> directly comparable: chronyd's are "spiky" while ntpd's are much
> flatter. I wonder if there is internal smoothing in ntpd that isn't in
> chrony. The kernel loop variables also behave differently, again
> making comparison meaningless.
The ntpd peerstats logs contains are only measurements that passed the
clock filter, which can drop up to 7 consecutive measurements. chronyd
has different filters, which normally drop fewer measurements (but
there is weighting).
> Is there a source I can look at for how to do such a comparison and
> what statistics should be gathered to monitor timekeeping quality?
A meaningful comparison would be to calculate the offset from the four
timestamps in ntpd's rawstats log and compare it to the offset in
chronyd's measurements log.
That should allow you to compare the stability of the synchronization.
It won't say anything about accuracy.
> Ideally we would not need a GPS slaved machine to do the monitoring
> and comparison, but methods that assume such a machine might also be
> of interest if there is no other way to obtain the information.
> Currently we are thinking of ntpd with the statistics command on on a
> third machine, but don't really know how we would process the results.
I'm not sure how useful that would be. If you need to compare the
accuracy of the clocks synchronized by ntpd/chronyd, you need a more
accurate and independent time source, e.g. a GPS refclock.
--
Miroslav Lichvar
--
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.