[time-nuts] clock-block any need ?

[Attila Kinali]

On Fri, 28 Dec 2012 19:18:06 -0800
Hal Murray <hmurray at megapathdsl.net> wrote:

> 
> attila at kinali.ch said:
> > From the data ntp gives me in the networks i manage. I hardly get any
> > jitter number below 1ms, even with unloaded network and unloaded hosts. The
> > 200us comes from the "usual" rtt time measurements on PCs. 
> 
> What sort of networks are you talking about?  Are you synchronizing over LAN 
> or WAN?  Do you have a local refclock?

LAN, mostly two switches inbetween machines, no local refclock.

> On LANs with non-ancient PCs, it's easy to get round trip times under 200 
> usec.  That makes it hard to get jitter over 1 ms.  Or, if the jitter is that 
> bad, how can you call the network unloaded?
> 
> Here are a couple of ntpq printouts:

For me it usually looks like this:

# ntpq -pn
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
*192.168.0.100   192.33.96.102    2 u  348 1024  377    0.450    0.530   1.132



Some better, some worse.

On the other hand, it's enough when these machines are synced better
than to a fraction of a second. They don't do anything critical with
their relative timing. Thus i never bothered to decrease the jitter.


				Attila Kinali
-- 
There is no secret ingredient
         -- Po, Kung Fu Panda