[Nagiosplug-devel] Re: IPv6 and check_real

[Karl DeBisschop]

On Sat, 2003-08-09 at 02:14, Jeremy T. Bouse wrote:
> On Sat, Aug 09, 2003 at 01:04:59AM -0400, Karl DeBisschop wrote:
> > On Sat, 2003-08-09 at 00:33, Jeremy T. Bouse wrote:
> > > Karl,
> > > 
> > > 	If I'm not mistaken the check_real plugin should be using the
> > > AF-independent code as it calls my_tcp_connect() to make it's connection
> > > which has the AF-indep code behind it... I just don't have any Real
> > > servers running on IPv6 although I hear from talking with Real at Linux
> > > World Expo this week they've open sourced the Helix server under an OSI
> > > approved license...
> > 
> > I don't have any servers to test either - but if the connection code is
> > in place, then I'll go ahead and add the options for it and we can put
> > out a call for a test.
> > 
> 	Yeah, if you look my_tcp_connect() and my_udp_connect() are just an
> easy way to call my_connect() without having to specify the protocol as
> TCP or UDP respectively... The my_connect() is the actual function that
> makes the socket connection and it is what has been written to be
> AF-indep from the ground up... The problem is when we find plugins that
> don't use these calls... If the function calls trace back to
> my_connect() then they are fine it's the ones that have their socket
> calls within themselves... All network based plugins should include
> netutils.h and utils.h while non-network based should only need
> utils.h...

That is correct up to a point - the is_host function got moved into
netutils - there are a number of plugins that wrap system calls that do
network connections. All these now need netutils.h, even though the do
not link to netutils.o.

Personanly, I find this to be a confusing state of affairs and I'd
prefer that is_host be in utils if it could.

> 	I don't believe I had gotten around to including the -4/--use-ipv4 and
> -6/--use-ipv6 to all the network based plugins as I had only done a
> couple for PoC testing and then got sidetracked on some other issues
> elsewhere and hadn't gotten back to finishing it...
> 
> 	Regards,
> 	Jeremy
> 
> > --
> > Karl
> > 
> >