F10 - dIsabling IPv4 addressing

[Robert Moskowitz]

Wolfgang S. Rupprecht wrote:
> Robert Moskowitz <rgm at htt-consult.com> writes:
>   
>> When I save ifcfg-eth0, NetworkManager reports the network is down.
>> If I start eth0 in NetworkManager, I get IPv4 addresses.  If I use
>> ifdown to bring eth0 back down, then ifup, I don't get IPv4 addresses,
>> only my IPv6 global assignment.
>>
>> What do I have to do to get the desired behaviour?
>>     
>
> You'll be doing yourself a favor if you ditch NetworkMangler and use
> the traditional network infrastructure.  
>
>     chkconfig network on
>     chkconfig NetworkManager off
>
> (Then run system-config-network and make sure your interfaces are
> configured the way you want then.  Save and reboot.)
>   

Yeah. I was afraid of that. Of course that means for the wireless, I get 
to go the old wpa_supplicant route too. But I am quite an old hand at 
that also.

>     root at poblano # service network restart
>     Shutting down interface eth1:                              [  OK  ]
>     Shutting down loopback interface:                          [  OK  ]
>     Bringing up loopback interface:                            [  OK  ]
>     Bringing up interface eth1:                                [  OK  ]
>     root at poblano # 
>     root at poblano # ifconfig
>     eth1      Link encap:Ethernet  HWaddr XX:XX:XX:XX:XX:XX  
>               inet6 addr: 2001:5a8:4:7d0:2e0:XXff:feXX:8dXX/64 Scope:Global
>               inet6 addr: fe80::2e0:XXff:feXX:8dXX/64 Scope:Link
>               UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>               RX packets:23 errors:0 dropped:0 overruns:0 frame:0
>               TX packets:18 errors:0 dropped:0 overruns:0 carrier:0
>               collisions:0 txqueuelen:1000 
>               RX bytes:2514 (2.4 KiB)  TX bytes:3164 (3.0 KiB)
>               Interrupt:19 
>
>     lo        Link encap:Local Loopback  
>               inet addr:127.0.0.1  Mask:255.0.0.0
>               inet6 addr: ::1/128 Scope:Host
>               UP LOOPBACK RUNNING  MTU:16436  Metric:1
>               RX packets:4273 errors:0 dropped:0 overruns:0 frame:0
>               TX packets:4273 errors:0 dropped:0 overruns:0 carrier:0
>               collisions:0 txqueuelen:0 
>               RX bytes:2326924 (2.2 MiB)  TX bytes:2326924 (2.2 MiB)
>
> First off, you'll want to make sure your /etc/resolv.conf has some
> ipv6 nameservers listed.
>
> Now I doubt the system is capable of running as a pure ipv6 host
> unless someone put in the effort to flush out the last few ipv4
> dependencies.  We do have ~3 decades of ipv4 dependencies wired into
> the code. There are no doubt lots of programs that were never updated
> to use ipv6 sockets.
>   

You cannot turn off IPv4 for lo. At least that I have found. The kernel 
is hardwired for IPv4 and you would have to build your own to rip it 
out, and then you will probably break a lot. I figure 10 years after we 
successful transition MOST of the ipv4 dependencies will be moved to 
unloadable modules (maybe for the 3.8 kernel!)

And for all the talk about IPv6 ready, you are right there are a lot of 
things we depend on that only work over IPv4. VNC is an example. Only 
RealVNC SUPPOSEDLY works with IPv4. Maybe the work to switch from VNC to 
TightVNC for FC11 could include adding IPv6 support....

At the last IETF, there was further talk about redesigning the APIs to 
only present names to the apps and blocking them from getting IP 
addresses. And perhaps even 'jimmying' some lower layer parts to control 
bad behaviour! My HIP effort is one noted as one of the models for 
replacing what apps see about IPv4 or IPv6 addresses.


>> Oh, I am logged in as root, so I don't have to futz with permissions
>> to fiddle with the interface.
>>     
>
> Sigh.  

Yes, one of these days all this stuff will just work right for regular 
userids.

And that will be in the 4.9 kernel :>'