[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gpsd-dev] Clarifications needed for the time-service HOWTO

From: Hal Murray
Subject: Re: [gpsd-dev] Clarifications needed for the time-service HOWTO
Date: Mon, 21 Oct 2013 15:01:38 -0700

address@hidden said:
> The public documentation insists that Stratum 1 servers are supposed to
> have fixed and latency-compensated paths.  I totally believe you when you
> say people screw this up, but that doesn't seem like a good reason to ignore
> the specification.

I find that hard to understand.  I don't remember any documentation that says 
"Stratum 1 must have fixed and latency-compensated paths."  Some stratum 1 
clocks don't even have a path so that statement doesn't quite make sense.  
(My first try, I thought you were talking about network paths, and that made 
even less sense.)

Stratum 1 servers get their time via some non-NTP back door like GPS or 
WWV/WWVB or ...

Some sources, like WWV/WWVB have a well defined radio propagation mechanism.  
You can compensate for that.  Is it fixed?  Mumble.  How accurate do you want 
it?  WWV propagation changes from night to day as the ionosphere moves 
up/down.  I think it's a few ms.  If you have a good GPS setup you can 
measure and track it.

Stratum 1 servers are supposed to know how accurate their stratum 0 clock is. 
 That accuracy gets passed down the chain and increased by network delays.  A 
good stratum 2 server can be better than a stratum 1 server.

I think that accuracy is compiled into the refclock drivers.  I don't 
remember any discussion of how to set it.

Yes, the bottom line is that it's easy to setup a bogus stratum 1 server.  
Probably the easiest way is to get the fudge setting wrong on a gizmo that 
needs it.  :)

address@hidden said:
> Are Ask Bj=F8rn Hansen and his crew at the NTP Pool crew doing a good job
> of kicking bogus Stratum 1s off their list?  If so, that's a thing to note
> in the HOWTO.  If they're doing a terrible job, that would be an even more
> important thing to note. 

They do a reasonable job of kicking off broken servers.  I don't think they 
check for stratum.  Their version of good-enough is not very tight.  Their 
target is 10s to 100 milliseconds rather then sub millisecond.  Their test is 
more alive and sane rather than good.

These are my opinions.  I hate spam.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]