On Wed, Feb 06, 2002 at 09:21:06PM -0500, Paul M Foster wrote:
> On Wed, Feb 06, 2002 at 12:27:48AM -0500, Robert Haeckl wrote:
>
> > > ... The more severe
> > > problem is that the DNS code on the local machines appears to initially
> > > ignore host.conf and hosts, until the internet nameservers tells it to
> > > go away.
> > >
> > > Surely DNS isn't _supposed_ to work this way?
> > >
> > > Paul
> >
> > You might want to check /etc/nsswitch.conf for the lines:
> >
> > hosts: files dns
> > networks: files dns
> > services: files
> >
> > if you haven't already done so. It's my understanding that the resolver
> > looks here now to see how to resolve names. According to the man file,
> > if the nsswitch.conf file is missing, libc.so.6 uses dns by default to
> > resolve names.
> >
> > -Robert
> >
>
> Hmm. I'll try that. I was aware that there was an nsswitch.conf file,
> but I didn't really know what it was for. And the man pages for the
> other files involved (resolv.conf, hosts, etc.) don't make much mention
> of this.
This didn't prevent telnet for forcing a dialout. However, further
experimenting showed that rlogin does _not_ force a dialout. I glanced
very briefly at some of the RFCs and man pages. Apparently, rlogin came
to life later than telnet, but the man page for rlogin says that telnet
will replace it shortly. Can't really see there's much difference in the
two.
Seems a shame to have to resort to setting up a DHCP/DNS server just to
prevent a dialout on telnet. I was hoping to make the server headless
and use telnet or ssh to talk to it without a lot of fuss. I still can't
surf from a workstation on the network, but that's another story.
Paul
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 15:50:51 EDT