Re: [chrony-users] chrony switch to low stratum very slow

[Bill Unruh <unruh@xxxxxxxxxxxxxx>]

chrony has no idea that the server is unreachable. It does not know you
unplugged it. All it knows is that on thisinstance it did not receive a reply.
And that it is bad for timekeeping if you bounce around from one server to the
other (eg one could have ns accuracy while the next has second accuracy). So
once it has settled on a good one, it sticks with it until it is clear that it
no longer answering. If you want accuracy and reliability, buy yourself a
number of GPS receivers, and even then, your govt could decide to jam all GPS
in your vicinity.

William G. Unruh __| Canadian Institute for|____ Tel: +1(604)822-3273
Physics&Astronomy _|___ Advanced Research _|____ Fax: +1(604)822-5324
UBC, Vancouver,BC _|_ Program in Cosmology |____ unruh@xxxxxxxxxxxxxx
Canada V6T 1Z1 ____|____ and Gravity ______|_ www.theory.physics.ubc.ca/

On Thu, 8 Aug 2019, chuang213 wrote:

> Thanks for the advice and I'll play around based on your advice.
>
> As you can see from the log, the delayed switch caused the local
> clock's offset from about 10 us to 300 us. I think it's better to
> consider the factor of unreachable server.
>
> Frank Huang
>
> On Thu, Aug 8, 2019 at 12:20 AM Miroslav Lichvar <mlichvar@xxxxxxxxxx> wrote:
> > On Wed, Aug 07, 2019 at 12:02:50PM -0700, chuang213 wrote:
> > > 2019-08-07 00:38:35 172.19.40.23     2    -14.307      0.054
> > > -6.028e-06 N  5  2.563e-05  7.701e-06  2.349e-04  1.930e-05  2.297e-04
> > > 2019-08-07 01:10:33 192.168.10.9     3    -14.307      0.000
> > > -3.306e-04 N  3  2.376e-04  7.130e-09  4.031e-04  3.629e-05  2.183e-03
> >
> > > it seemed it took 16 seconds (the poll period) to switch server from
> > > stratum 3 to stratum 2. However, it took almost 32 minute, or 1918
> > > seconds to switch server from stratum 3 to stratum 2.
> >
> > To me that looks about right. The second source not only has a higher
> > stratum, but it has also has a larger root delay and dispersion. With
> > the default stratumweight and reselectdist, the lower stratum source
> > has to gain about 1300 milliseconds to be comparable to the other
> > source, and if the dispersion rate is at 1 ppm (maxclockerror), it
> > will take more than 1300 seconds for the scales to switch in favor of
> > the other source.
> >
> > You can set stratumweight and reselectdist to 0, but there will still
> > be some delay due to the increase in root delay and dispersion, and
> > the reselect process itself.
> >
> > > Although I know lower stratum server is in favor but when it is
> > > unreachable, it should switch to other reachable server, sometimes
> > > lower stratum.
> >
> > chronyd doesn't care about reachability very much. An old accurate
> > source that tracked very well may still be preferred over less
> > accurate/stable sources.
> >
> > --
> > 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.
>
> --
> 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.

--
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.