[ntp:questions] Re: NTP clients not syncing up to servers?

David Woolley david at djwhome.demon.co.uk
Mon Oct 10 06:55:47 UTC 2005


In article <43496015.80409 at ntp.isc.org>, mayer at ntp.isc.org (Danny Mayer) wrote:
> Ted Beatie wrote:

> Based on the above the internal NTP server has a stratum of 2 and will 

W32Time always claims this stratum!  Some versions of the Free BSD SNTP 
software also did so.

> > dstport=123, leap=00, stratum=2, precision=-7, rootdelay=0.000,
                                  ^                          ^^^^^
rootdelay shouldn't normally be zero for a stratum two server
(although it could be if you fudged the stratum on a server with
a reference clock).

> > rootdispersion=9733.109, refid=bigbird.babcockbrown.com, reach=377,
                   ^^^^^^^^
This is the reject reason.  In strict NTP, this means that the server has
been unsychronised for so long that it's time is not known to better than
+/- 9.7 seconds.  The refid indicates that it isn't using a fudged clock,
so the zero rootdelay is an implementation error.  Large (or possibly
reflected) root dispersions are another W32Time indicator.

> > filtoffset= -1572.7 -1562.2 -1567.0 -1572.8 -1562.8 -1567.7 -1573.4 -1563.1,

> This appears to indicate it received just one packet which is not enough 

No it shows at least 8 packets.

My guess is that the internal server isn't running NTP but rather is
running "Seattle" Network Time Protocol, Microsoft's invalid implementation
of SNTP, otherwise known as W32Time.  This should never be used to serve
time to anything but a leaf node and should never be used as a source for
a node that has access to a real NTP server.

> to synchronize anything. How long did you wait for the server after it 
> was started to interrogate this server? You need to wait at least 15-20 

I don't think we have the figures for this server, but the upstream server
was running for over 4 hours judging by its reftime.




More information about the questions mailing list