Harlan Stenn wrote:
The best reason to have your S2 servers chat with different vendors of
different technologies of refclocks is that sometimes they can go insane.
It's one thing for the GPS system to go insane.
It's something else for a vendor to discover a bug in one of their GPS
products. The odds are good that different GPS vendors will not have that
same bug.
I don't mean to pick on GPS here - feel free to choose whatever technology
you prefer for that "slot".
This is exactly the reason why I am using multiple technologies as S1
servers in our corporate network:
I started out with three different 'black box' NTP/GPS servers, from two
different vendors, plus a home-made Motorola Oncore UT+ hosted on a
FreeBSD machine.
Over the next ~5 years all the 'black box' systems failed badly at least
once: I.e. they became falsetickers, claiming that the current time was
something like 2034. :-(
Since the Oncore didn't do this, I've based the new infrastructure on
4-5 of the last available Oncore's, plus one Garmin OEM 18 VLC, a
Meinberg DFC-77 radio receiver located in a German office, the old boxes
as backups, and a few internet peers as the last resort.
With same-LAN location as a S1 server, minpoll = maxpoll = 16 seconds,
I'm seeing ~100 us precision on secondary servers.
Terje
--
- <[EMAIL PROTECTED]>
"almost all programming can be viewed as an exercise in caching"
_______________________________________________
questions mailing list
[email protected]
https://lists.ntp.isc.org/mailman/listinfo/questions
