On Sun, 8 Oct 2000, Chris Pinnock wrote:

> I've been seeing a lot of this lately:
> 
> Oct  7 16:29:33 baer xntpd[102]: synchronized to 138.37.80.10, stratum=3
> Oct  7 16:29:32 baer xntpd[102]: time reset (step) -1.075148 s
> Oct  7 16:29:32 baer xntpd[102]: synchronisation lost
> Oct  7 16:34:32 baer xntpd[102]: synchronized to 138.37.80.30, stratum=2
> Oct  7 16:34:34 baer xntpd[102]: time reset (step) 1.032995 s
> Oct  7 16:34:34 baer xntpd[102]: synchronisation lost
> Oct  7 16:39:10 baer xntpd[102]: synchronized to 138.37.80.10, stratum=3
> Oct  7 16:39:09 baer xntpd[102]: time reset (step) -1.083195 s
> Oct  7 16:39:09 baer xntpd[102]: synchronisation lost
> 
> 
> I don't recall this being a problem on 1.4.2 (perhaps I'm wrong).

You seem to be "clock hopping". Better to winnow down to one or two of
your closest and most stable reference clocks. Getting time from a
close stratum 3 and a distant stratum 2 is especially screwed up; a
few at the same stratum that stay close together, themselves, is OK.

BTW, Dave Mills responds to xntpd bug reports with "That software is
buggy, and no longer supported". ntp-4.x (the package) seems to work
fine with NetBSD-1.4.3; NTP4 is already in -current and NetBSD-1.5.



Frederick