[ntp:questions] Re: ntp4 new functionality testing

David L. Mills mills at udel.edu
Sun Feb 20 16:20:45 UTC 2005


We support two testbeds here, each including 6-8 systems of various 
manufacture on dedicated subnets. We have scripts that build the public 
software with several common options - with and without reference 
clocks, debugging, cryptographic authentication, etc. The daemon can be 
operated in siumulation mode where the basic protocol module and 
algorithms are driven by synthetic network jitter and oscillator wander. 
We use that and simple test scripts to test convergence and stability in 
various scenarios with very large time and frequency errors, for example.

We test things like Manycast, Autokey and selected reference clock 
drivers using ad-hoc scripts and local radios, but have no formal test 
suite. It would be a good idea to have a comprehensive suite of test 
scripts, but there are a great many options and configuration schemes 
and not enough time or enthusiasm among the volunteer Corps to construct 
scripts for all of them.

Much of the testing, especially recently when folks have pushed the 
envelope with unexpected parameter extremes, is prompted by some 
complaint or other about behavior the program was not intended or 
designed for. It's hard to anticipate such scenarios and design test 
scripts for these cases. As you can see, a thorough test suite is a 
serious project in its own right.


Sajitha T T wrote:
> Hi,
> Is there any test suite/cases available with which we can test the ntp4 
> new functionality, specifically autokey authentication, automatic 
> configuration, ipv6 etc.  Greatly appreciate any pointers on this.
> Regards,
> Sajitha

More information about the questions mailing list