Am Sonntag, 30. Dezember 2012 schrieb Ludovic Courtès:
> This is because this test relies on the presence of /etc/services when
> looking up the “telnet” service by name (which is missing in chroots.)
> Possibilities include:
>   • patching the test avoid that (for instance, by changing ‘localhost’
>     to ‘127.0.0.1’, ‘telnet’ by ‘23’, etc.), unless it would no longer
>     make sense;
>   • patching the test machinery to skip the test;
>   • turning off all the tests, as a last resort, if they are all really
>     network-dependent.
> Any idea what would be the most relevant approach here?

I think both of the first two approaches would work, while the third one 
should be avoided.

Andreas