[time-nuts] NTP - PPS

Paul tic-toc at bodosom.net
Sun Nov 9 23:34:46 UTC 2014


On Sun, Nov 9, 2014 at 5:37 PM, "Björn Gabrielsson" <bg at lysator.liu.se>
wrote:

> Nevertheless I was a bit surprised that the computer continued to serve
> time claiming to be a S1-server,
>

Enabling PPS behaves a bit like a back-door in the recent/current
releases.  Among other somewhat surprising behavior the code will use the
PPS signal until the "prefer" clock delta is greater than 400ms.  Naturally
this  can couple in unfortunate ways.

The moral of this story is follow best practice and connect the validity
"flag" associated with your PPS source to NTPd.



More information about the Time-nuts_lists.febo.com mailing list