[ntp:questions] Re: offset of two acts servers

Richard B. Gilbert rgilbert88 at comcast.net
Fri Feb 3 13:23:24 UTC 2006


Garrett Wollman wrote:

>In article <11u59933rt89sae at corp.supernews.com>,
>rtxo  <gnu at wraith.sf.ca.us> wrote:
>  
>
>>Oddly, one of the acts hosts is consistently 3-5 milliseconds behind,
>>and the other 3-5 milliseconds ahead:
>>    
>>
>[...]
>  
>
>>I wonder whats going on with these hosts. The refclock_acts driver code has
>>a comment that claims the acts adjusts for the round trip delay.
>>    
>>
>
>Not sure how relevant it is, but have a look at
><http://khavrinen.csail.mit.edu/ca-ntp-offsets.pdf>.  (It's more
>interesting on slower machines.)  That's offset on the Y axis and MJD
>on the X axis.  Note the odd behavior of time-b.nist.gov (another ACTS
>server).
>
>-GAWollman
>
>  
>
Interesting, but I think it shows more about the network connections 
between server and client than the quality of the servers.  For example, 
NAVOBS1, as received here (~10 miles southeast of the Philadelphia 
Airport) frequently develops an offset of 40 or 50 milliseconds during 
the hours 0500 - 1200 UTC.  Time-b.nist.gov is all over the map during 
daylight hours in the US.  OTOH, I use a server in NYC that shows pretty 
much a flat line at an offset of +2ms all day every day! 




More information about the questions mailing list