[ntp:questions] Trying to use Dimension 4 time keeper
david-taylor at blueyonder.co.uk.invalid
Sat Sep 14 14:59:40 UTC 2013
On 14/09/2013 14:56, David Woolley wrote:
> Offset doesn't tell you this. It tells you the difference between the
> best estimate of the local clock and best estimate of the time on the
> immediately upstream server, at the time of the last poll. The latter,
> in particular, may be subject to systematic errors (in most internet
> cases, these will dominate) and both are subject to random errors.
> Generally the non-systematic component of the error in the local clock
> will be significantly less than the RMS offset.
> Jitter is probably a better estimate of quality with respect to random
> errors. There is no way of measuring systematic errors, except by
> accessing a source known to have lower ones.
Yes, here I'm typically measuring against local stratum-1 servers. I'll
take your word on offset, but I've found that the lower the offset,
typically the lower the jitter, and the better the performance.
The output from e.g. ntpq -pn and nptq -crv is always something I
like to see when trying to diagnose problems.
More information about the questions