[ntp:hackers] Modern FreeBSD NTP Stratum 1 server?

Hal Murray hmurray at megapathdsl.net
Thu Dec 17 20:10:49 UTC 2015


karl at denninger.net said:
> If others have good PPS service on FreeBSD 10.2 and NMEA the problem might
> well be the hardware :) 

There are two potential problem areas that I know about in the hardware area.

The first is that many PPS signals get wired up directly without going 
through the typical RS-232 level converter.  I haven't seen that fail.  I 
wouldn't be surprised to see extra pulses on long noisy lines.  It's more 
likely to fail if your PPS pulse is 0-3V as compared to the older 0-5V.

The other is that many PPS pulses are narrow, typically 10 microseconds.  
Some systems don't catch them.  The clockstats info for the ATOM driver  has 
counters for this.

Grumph.   Looks like I didn't get the documentation for that into the system.
After the normal header of date, time, and ipaddress, there are 5 counters.
  The first is the total number of PPS pulses the system has seen.
  The second is the number of PPS pulses that the ATOM driver has processed 
in the last polling interval.
  The next two counters are error conditions.
  The last counter is the number of times that the ATOM driver looked for the 
data on the latest pulse and there wasn't any new info available.

57373 72413.206 127.127.22.1 1357868 35 0 0 29
57373 72477.207 127.127.22.1 1357909 41 0 0 23

So that setup is missing lots of pulses, but getting enough to keep the 
system happy.


-- 
These are my opinions.  I hate spam.





More information about the hackers mailing list