Re: [hobbit] silly-a** question on hobbit, timezones, and the dread purple status

[henrik (at) hswn.dk (Henrik Stoerner)]

On Wed, Mar 08, 2006 at 12:20:51PM -0600, Larry Barber wrote:

> On 3/8/06, Kauffman, Tom <KauffmanT (at) nibco.com> wrote:
> >
> > My hobbit server is in Indiana (US) and we've just decided to start
> > doing the 'change the clock twice a year' shtick again. Do I need to
> > worry about what alerts we'll get when the time shifts by an hour, twice
> > a year? Or does hobbit walk the status messages and re-date them when
> > the shift occurs?
>
> The purple status relies on the mod time of the status file, which uses the
> Unix sytem time of seconds since the "epoch" so it shouldn't be affected by
> the time zone or daylight savings time.

That's not *quite* right, but the end result is the same.

Hobbit doesn't maintain any kind of status in files the way BB does.
What Hobbit does is to keep a timestamp in memory - but the effect 
is the same. All of these timestamps are in the Unix "epoch" format,
and therefore they are independent of what the local time may be. 

(This is definitely one thing the Unix designers got right - storing
time in a format which depends on your location, and then trying to
compare them across multiple systems, is bound to fail. The only
sane way of doing that is to use a location-independent counter, and
the Unix time_t epoch thing is exactly that. The POSIX guys then
goofed badly on the issue of leap seconds, but we'll have to live
with that).


Regards,
Henrik