[ntp:questions] question regarding NTP configuration for clusters, and "cluster time" stability
unruh-spam at physics.ubc.ca
Sat Nov 21 03:38:28 UTC 2009
"Richard B. Gilbert" <rgilbert88 at comcast.net> writes:
>E-Mail Sent to this address will be added to the BlackLists wrote:
>> rotordyn at yahoo.com wrote:
>>> Look at it this way: We produce a system (in the form
>>> of a cluster) that works today, but that can drift away
>>> from UTC since it currently doesn't accept any external
>>> time reference. That drift isn't a huge issue, but over
>>> the multi-year lifespan of the hardware, can be significant.
>>> If we allow the use of external NTP servers, we now have
>>> opened up what was a closed system, and must prevent that
>>> new input source from causing instability. ...
>>> An alternate approach would be to make our cluster
>>> software resilient to intra-node time variations, but it
>>> was deemed simpler to use NTP to have the cluster accept
>>> an external UTC source.
>There are solutions that do not require that you get time from the internet!
>A GPS timing receiver, for example, requires only an unobstructed view
>of most of the sky. The Pulse Per Second (PPS) output is generally
>accurate to within +/- 50 nanoseconds. You lose a great deal of
Well, no, on the cheapest option, the GPS18xLVC it is about 500ns, and
furthermore, the computer almost certainly cannot respond to the interrupt on a
50ns basis-- tests I did suggest more like 1-2usec (or worse if it happens to come
in during a time when interrupts are switched off).
>accuracy getting the time into your computer. Determining just how much
>you lose is difficult but you can assume that you know the time to
>within, say, 100 microseconds.
You can reliably get a few microseconds 1 or 2)
>If you connect the GPS to a computer running NTPD, the machine becomes a
>stratum one server, meaning that the computer is getting time directly
>from an atomic clock on board one of the NAVSTAR satellites. As of the
>last time I heard there were 27 satellites in orbit and your GPS
>receiver should be able see at least four and be able to determine your
>latitude, longitude, and elevation to within about 50 feet and the time
>to within 100 nanoseconds.
The receiver may. If you tell it where you are, then it can get the time with many
>A shortwave radio receiver should be able to pick up a time signal from
>WWV, the NIST HF broadcast from Fort Collins, CO (2.5MHz, 5MHz, 10MHz,
>20MHz, . . . . There are special purpose receivers that will
>automatically select the best radio frequency from the five or six
>frequencies that NIST uses and decode the timing signal. WWVH
>broadcasts from Hawaii.
Since the delaty is many milliseconds ( not microseconds) and variation in the
delay is many microseconds, certainly a second best option.
>A VLF receiver should be able to pick up a signal from WWVB.
>See the NIST web site for more information.
>Other countries offer similar services.
More information about the questions