18 sec

[Garrett Wollman]

<<On Tue, 29 Mar 1994 10:12:19 +0200 (MESZ), Hinrich Eilts <eilts at late.e-technik.uni-erlangen.de> said:

> I synchronized FreeBSD-1.1-BETA to a stratum-1 timeserver by ntpdate
> and xntpd and compared localtime by "date" with it and two other
> machines, which are synchronized by xntpd too.  FreeBSD has a
> difference of -18 secounds to all other machines (a NeXT-Station
> with NeXTStep, a Sun with SunOS 4.x and a HP9000-750 with
> HPUX).

What you are seeing is the interaction of several different `features'
of NTP and FreeBSD.  I'll try to explain, and again hope that somebody
out there can provide a complete solution.  Here are the problems...

1) The timezone database does not reflect the fact that the first time
step, in going from GMT (pre-1972) to UTC (1972 to present) was
actually 10 seconds, and not just one like all subsequent leap
seconds.

2) The timezone database DOES include all subsequent time steps.

3) The NTP code ``handles'' leap seconds by turning the clock
back when they happen.  By contrast, the timezone code assumes that
time_t's are increasing (which they ought to be; I hold that xntpd's
behavior is simply wrong).  To wit:

NTP code wants:		Real Time	UNIX Time
---------------		---------	---------
30 June 1994, 23:59:59	t		s
30 June 1994, 23:59:59	t+1		s
01 July 1994, 00:00:00	t+2		s+1

Timezone code wants:	Real Time	UNIX Time
--------------------	---------	---------
30 June 1994, 23:59:59	t		s
30 June 1994, 23:59:60	t+1		s+1
01 July 1994, 00:00:00	t+2		s+2

A temporary (ugly hack) solution is to use `posix/your/time/zone'
rather than `right/your/time/zone' (which is the same as
`your/time/zone') for /etc/localtime.

-GAWollman

--
Garrett A. Wollman   | Shashish is simple, it's discreet, it's brief. ... 
wollman at lcs.mit.edu  | Shashish is the bonding of hearts in spite of distance.
formerly known as    | It is a bond more powerful than absence.  We like people
wollman at emba.uvm.edu | who like Shashish.  - Claude McKenzie + Florent Vollant