On Mon, Aug 14, 2017 at 05:51:26AM +0000, Jahagirdar, Bhushan wrote:
> Hi Miroslav,
> 
> Attaching the log files.
> Please let me know if you need more data to analyze the use case.

We need to see also the refclocks log (enabled by log refclocks in
chrony.conf).

I suspect the problem happens when the refclock is stepped right in
the middle of the 4-second polling interval, so there are two samples
before step and two samples after step. The median filter can't handle
that well and the dispersion jumps to a large value, which then takes
a long time to settle down. As a workaround, I think you could add
"filter 1" to the refclock log, which will prevent the dispersion from
changing. As others said, this might not be the best way to test
things. Reference clocks are not supposed to jump by months.

-- 
Miroslav Lichvar

-- 
To unsubscribe email chrony-users-requ...@chrony.tuxfamily.org 
with "unsubscribe" in the subject.
For help email chrony-users-requ...@chrony.tuxfamily.org 
with "help" in the subject.
Trouble?  Email listmas...@chrony.tuxfamily.org.

Reply via email to