Bug#686970: Inconsistent ipv6 handling of localhost in netbase and netcfg

[Bjørn Mork <bjorn@mork.no>]

Philipp Kern <pkern@debian.org> writes:
> On Mon, Sep 17, 2012 at 11:30:58PM +0200, Marco d'Itri wrote:
>> On Sep 17, Philipp Kern <pkern@debian.org> wrote:
>> > True, but somehow I feel that freeze time is not the time to change this.
>> I disagree, netcfg should be fixed. There is plenty of time to look at 
>> any fallback, and even reverting the change if needed.
>> This is 2012, Europe and Asia are out of IPv4 networks and we cannot 
>> afford substandard IPv6 support anymore.
>
> It affects services listening on localhost.

Well, those services are already affected in any case. Any bug triggered
by cleaning this up would already be triggered depending on DNS
availability.

/etc/hosts is not the primary source for names, and if you put anything
there then it should really, really not be some locally invented
mumbo-jumbo name.  That just creates inconsistencies between the
connected and unconnected system state and is not helpful at all.

For example:

    bjorn@canardo:~$ grep localhost /etc/hosts
    127.0.0.1       localhost
    ::1     ip6-localhost ip6-loopback

    bjorn@canardo:~$ host localhost
    localhost has address 127.0.0.1
    localhost has IPv6 address ::1
    bjorn@canardo:~$ host ::1
    1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.ip6.arpa domain name pointer localhost.

This is not good...  Which addresses will a service listening on
"localhost" listen too?  Seems it depends on whether DNS is available or
not.  Isn't that seriously buggy?


Bjørn