Re: [opensuse] chrony and hwclock

Roger Oberholtzer said the following on 07/15/2011 07:17 AM:
That seems to be the point of chrony. Access to the time source need not
be regular. Obviously the longer it is not connected the more time will
suffer.

Isn't this what NTP does?
Only with 'layers' (?'stratum'?) of authority for fall-back.


I guess it assumes that the error in the real time clock is
basically linear.

That's how I read the way NTP deals with 'drift'.


To the extent this is true, it could perhaps do a good
job. I worry that the error in the rtc may not be linear, especially if
(1) the system is disconnected from the power mains and runs on the PC
battery,

There are published studies on this that I've seen but not
studied/bookmarked, that seem to indicate

a) drift is very low on modern devices
b) it is linear to the first order.

I'd go google If I were you.

or (2) temperature changes drastically (hot in summer and cold
in winter).

That depends on the environment, obviously.
Outside the rated temperature for semiconductors ... who knows.

Here in the GWN I once had a project where we worked in Port-a-Cabins
that were sub-zero (Celsius) overnight. We had to turn on the heaters
and go get "breakfast" for an hour each morning while they warmed up.

But that's commercial grade stuff.
I've also worked with MIL-SPEC and AUTO-SPEC.

Outside of military, an automobile is about the worse environment for
electronics you will encounter. Its bad enough to be the electronics
in, say a traffic light controller cabinet, exposed to heat
fluctuations. But auto electronics has to deal with that and more.
Vibration, salt, dirt ....

Yes, I've worked with aviation electronics too, which is more extreme,
but just about everyone encounters automobiles whereas few of us have to
deal with the conditions at 60,000 feet.


The point I am trying to make is that the chip and board makers *DO*
know about these conditions and *DO* make components that are stable
under them.

If you choose to use chips & boards intended for a home theatre under
these conditions, then yes there will be fluctuations. If you choose
the suitable - and probably more expensive - components that are
designed for that environment then the problems you describe will not be
apparent.


Unless you are using inappropriate hardware, the advantages you claim
for chrony exist in NTP. Are there some advantages of chrony that NTP
doesn't have which you have not mentioned?

--
Faith may be defined briefly as an illogical belief in the occurrence
of the improbable...A man full of faith is simply one who has lost (or
never had) the capacity for clear and realistic thought. He is not a
mere ass: he is actually ill. -- H. L. Mencken
--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse+help@xxxxxxxxxxxx

Relevant Pages

  • Re: NTP vs chrony comparison (Was: oscillations in ntp clock synchronization)
    ... the time constant could be reduced to anything you want using ... a type-II linear phase-lock loop, the characteristics of which are very ... not be strictly linear. ... The NTP discipline is basically a type-II feedback control system. ...
    (comp.protocols.time.ntp)
  • Re: Need help setting up NTP in simple network
    ... > environment (Sun Studio or gcc and gmake). ... > Sunfreeware and Blastwave take all the effort out of it! ... I appreciate the info on the newer version of NTP, but actually, I work ...
    (comp.sys.sun.admin)
  • Re: Server offset included in served time?
    ... Bill has found that in his environment, the algorithms used by chrony preform better than those used by NTP. ... My recent experiments show that the same algorithms that are great in NTP for the long haul, actually hinder its ability to get the offset and frequency right under many typical startup conditions. ...
    (comp.protocols.time.ntp)