Re: monit/protocols ntp3.c

[Martin Pala]

Jan-Henrik Haukeland wrote:
> On Jan 28, 2005, at 22:18, Martin Pala wrote:
>
> > Jan-Henrik Haukeland wrote:
> >
> > > On Jan 28, 2005, at 18:55, Martin Pala wrote:
> > >
> > > > Log message:
> > > >     Add Michel Marti's NTP protocol test.
> > >
> > > Hmm, we should really have fixed the check_udp_socket() first and 
> > > change the patch and remove the code for *not* checking if a udp 
> > > socket was ready. I.e. fix the root cause and not allow the 
> > > workaround. Anyway, now that it is checked-in we'll just have to fix 
> > > the patch-code afterwards when we have a proper check_upd_socket 
> > > implementation.
> >
> >
> > I forgot to mention that i omited the workaround introduced in the 
> > patch (current monit check_udp_socket behavior is kept even with ntp3 
> > patch integration).
>
>
> Thanks for the explanation, I thought for a moment I was in the twilight 
> zone :-) Good decision and the test will probably work anyway? We can 
> fix check_upd_socket later.

Yes - the test works without problems, there is only one extra datagram 
(1 byte) generated in check_udp_socket() before the specific test (it 
just checks that the port is accessible, i think we can live with it ... ;)