Re: hard crash on leap second
On Thu, Jan 01, 2009 at 07:44:07AM -0600, John Hasler wrote:
> I wrote:
> > Debian has time servers:
> > 0.debian.pool.ntp.org
> > 1.debian.pool.ntp.org
> > 2.debian.pool.ntp.org
> > 3.debian.pool.ntp.org
> I prefer to use reliable sources. Also, as Chrony maintainer I feel that I
> should use my default configuration as much as possible.
> > It seems that ntp4.ja.net is preferred to ntp0.zen.co.uk despite the
> > longer delay, presumably because it's a stratum 1 server?
> Look at the jitter. The algorithm considers more than just the stratum and
> the round-trip delay.
> John Hasler
You and I appear to be the only people posting to this thread who use
chrony. I run it just the way you set it up in the package. (except
that I'm always connected to the Internet, and edited out the
'offline' from the server spec.s) Because of this long thread, and
especially because of reported crashes, I looked at my chrony logs.
I saw nothing remarkable, except that there was no evidence of a one
second leap. Because of the pool, there was not, in my record,
frequent data from a single ntp server, so I'm not sure I could have
seen a leap spread across four ntp servers. Did you see any evidence
of the leap?
Not that it really matters. I'm a physicist who is comfortable with
the fact of Einsteinian special relativity. For me its important that
my clock run smoothly, and that my software not crash when it discovers
a discrepancy between two clocks at widely separated locations. And,
the idea of leap seconds is something of an intellectual abomination,
So, thanks for making chrony work the way a clock should work.
Paul E Condon