Roger Oberholtzer wrote:
On Fri, 2011-07-15 at 13:41 +0200, Per Jessen wrote:
Probably, but not necessarily. The local clock _could_ be quite accurate by itself. I use ntp with a dcf77 receiver because it's easy and accurate, but I don't actually need the accuracy to more than 1 second, and only between my own machines.
One second accuracy for me would be a disaster. We need to locate things within a decimeter. At 90 km/h, that is 1/250 of a second. So we would like time to be at least 1/500 of a second accurate. By 'accurate' I mean between the systems (the GPS receiver being one of them), not absolute.
2ms accuracy - with ntp that's not a problem, but it seems to be asking a lot when you do not have a stable time source?
I define 'seldom' as being when the GPS is on and has a fix. For example, if our vehicle is turned on in a garage, the GPS may not provide PPS signals and a GPS-based time. It too is providing an internal time maintained between satellite access.
So now "seldom" is directly related to how often the vehicle is active? Are we talking a moon buggy or a city tram? :-)
Depending on the season. a bit of both.
I shouldn't have asked. :-)
In my laptop case with Network Manager, it moves to localhost at boot because there is no other source available. There is one configured - but it is not available at that time. The problem is that it seems not to re-evaluate the decision. So, when the NTP server is eventually available, it does not care as it is using localhost. So, I run 'rcntp restart' and all is well again.
Here is an excerp of my ntp-log, you'll see it changing servers/stratum several times per day: http://files.jessen.ch/ntp-sync LOCAL is localhost, GENERIC is DCF77, the two ip-addresses are a stratum 0 and a stratum 1 server respectively. -- Per Jessen, Zürich (20.6°C) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org