[ntp:questions] Re: ntpdate steps 86400 seconds

David Woolley david at djwhome.demon.co.uk
Sat Oct 1 22:25:38 UTC 2005


In article <433e7dab$0$67262$157c6196 at dreader2.cybercity.dk>, 
khanh.d.pham at gmail.com wrote:

> reference time:    c6e84f4c.00000000  Fri, Sep 30 2005 23:57:00.000

That is suspicious.  A reference clock with sufficient accuracy to
be a valid source for NTP is unlikely to be read exactly on the
second.  A possible indication of broken server software, or inadequate
clock hardware.

> reference time:    c6e9a144.00000000  Sat, Oct  1 2005 23:59:00.000
> originate timestamp: c6e9a144.ea4521cf  Sat, Oct  1 2005 23:59:00.915

Both these timestamps are set by the server.  The *server* is broken!
As it is clearly not running ntpd (suspiciously precise reference times)
this is not an ntpd or ntpdate problem.  I can't remember if you told
us who supplied the server, but, if you did, and they don't own up
here, you will have to take this up with them directly.

At a guess, the server is using a broken down time format, not the
second count, and is reading the date part after the time part.
I'd expect ntpd to read the time directly as seconds from an epoch,
and not suffer from this problem




More information about the questions mailing list