[ntp:hackers] odd behaviour (one more time)

David L. Mills mills at udel.edu
Mon Sep 8 17:10:32 UTC 2008


Your attachment got stripped of again.

I am running p127 in eight machines here with no such symptoms at all. 
As I said earlier, I even torqued one of them some tens of milliseconds 
and watched it recover with the intended transient response. At the 
moment, all except the ACTS modem server are within 100 ns and show no 
instability or oscillatory behavior. The ACTS modem shows offset of 
several milliseconds, but it has a poll interval of 36 h. All machines 
are either Solaris or FreeBSD and have a variety of reference clocks and 
secondary servers.

So, I have no useful insight.


Reg Clemens wrote:

> [[ Im sending this message a second time, since the plot got
> stripped off of it the first time around ]]
> This is 4.2.5p127, running since early yesterday.
> There was a -3ms offset in the loopstats when it
> started up, and it has been oscillating ever since,
> starting with an amplitude of 35ms and period of
> 3h20m (or so) with the amplitude SLOWLY decreasing.
> This is not good, but I dont see how I could have caused it.
> A plot of the current 24hr period is at
> <http://www.swcp.com:/~reg/offset.jpg>
> Thoughts?

More information about the hackers mailing list