[ntp:questions] Re: Can sychronize with a slightyly different NTP versios

David Woolley david at djwhome.demon.co.uk
Fri Mar 3 22:27:00 UTC 2006


In article <8oudnUzhLp1PNJXZRVn-iw at comcast.com>,
Richard B. Gilbert <rgilbert88 at comcast.net> wrote:

> > server 127.127.1.0  burst iburst

> Try:
> server 127.127.1.0              # Local clock, unit 0
> fudge 127.127.1.0 stratum 10

> I believe that what's happening is that without the "fudge" statement, 
> the client sees the server as "stratum 16" which means "unsynchronized" 

Without the fudge statement it sees it as stratum 5 (which you see from 
the ntpq output).  The reason for always fudging this is that 5 is too 
close to the root and there is too great a risk of a client with 
access to valid ntp severs using the time from this bogus server.

You were correct to drop burst and iburst.  I don't believe either
of them do anything for reference clocks and one SHOULD NOT use burst
unless one really understands the implications.

The flash code that was reported (0x020) means that either the leap
bits are set to the alarm state or the stratum is 16.  Normally, in this
configuration, that should only happen if the server was polled too soon.

I'm not quite sure whether kiss of death can be sent without being
explicitly enabled, but the burst keyword on the client may well 
trigger one and one of the indicators of KoD is the leap bits being
set to alarm.  It might be worth trying removing all the bursts.

"kiss of death" means that the server is telling the client to deconfigure
the server because it is polling the server abusively frequently.

> It would probably be a lot easier and consume far fewer resources to use 
>   rdate in a cron job.

That depends on what level of phase and freqeuncy error are tolerable.




More information about the questions mailing list