[ntp:questions] Is a high-stratum falseticker possible? Complex configuration and unexpected result

[Brian Utterback]

On 04/26/12 06:39, rumegauu at ugpa.ru wrote:
> Hello,
> I designed a complex configuration for a huge intranet network some
> time ago. The network  is completely isolated from the Internet and
> uses internal time sources. Circuits aren't any good there, so I
> decided to use some peer-to-peer synchronization in addition to the
> main hierarchy and enabled tos orphan.
>
> I hardly need to say that most of the hosts are embedded devices which
> aren’t easily accessible. I've built ntp4.2.6 from the tarball and
> deployed it across the devices. Various simulations were passed with
> success before the launch in 2011. It has been running well for 6
> months until some unexpected troubles come.
>
> Could you explain to me HOW this is possible? The prime server with
> stratum 6 declared falseticker, when the peer with stratum 12 is
> considered to be true.

the determination of falseticker is done without regard to selection by 
stratum. All of the reachable servers are used first to determine the 
limits of the true time, and any servers outside of those limits are 
marked as falsetickers. That is one reason why you want to have as many 
servers as possible be independent of each other, otherwise one server 
could influence the others and overrule the good ones. In this case 
however, your falsetickers are seconds away from each other, which is 
why they are marked as falsetickers.

-- 
blu

Always code as if the guy who ends up maintaining your code will be a
violent psychopath who knows where you live. - Martin Golding
-----------------------------------------------------------------------|
Brian Utterback - Solaris RPE, Oracle Corporation.
Ph:603-262-3916, Em:brian.utterback at oracle.com