On Sun, Apr 03, 2005 at 03:15:23PM +0200, Thomas Hood wrote: > I am reading RFC 3927 "Dynamic Configuration of IPv4 Link-Local > Addresses" and I notice that it says. > > > IPv4 Link-Local addresses should therefore only be used where stable, > > routable addresses are not available (such as on ad hoc or isolated > > networks) or in controlled situations where these limitations and > > their impact on applications are understood and accepted. This > > document does not recommend that IPv4 Link-Local addresses and > > routable addresses be configured simultaneously on the same > > interface. >
Later on, section 1.9, has a 'SHOULD NOT' (as per RFC2119) and
throughout the document it explains how to cope/handle the case of
multiple addresses assigned to an interface.
Just as in IPv6 where both globally routable and link-local addresses
exist on an interface simultaneouly, I don't perceive any particular
issue with having multiple IPv4 address per interface.
IPv4LL (LL = Link-local) addresses and IPv4 global address
> Given this, I am beginning to doubt that the zeroconf package should
> install a hook script in /etc/network/if-up.d/ which runs
>
> /usr/sbin/zeroconf -i $IFACE
>
> at a time when ifup has already configured the interface following some
> method or other, selected in /etc/network/interfaces. I don't think
> that it is appropriate that I end up with multiple addresses assigned to
> the same interface:
It is a normaly, natural consequence in IPv6 and is also possible when
using IPv4 anycast addresses as well. IPv4LL address might be the first
time most users encounter multiple addresses per-interface though.
> # ip addr show ethp_0
> 8: ethp_0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
> link/ether 00:80:c7:ee:88:d6 brd ff:ff:ff:ff:ff:ff
> inet 192.168.1.8/24 brd 192.168.1.255 scope link ethp_0
^^^^
/
That is a concern, that address -
should really have global scope.
Was this address alocated via zeroconf? If so, that is a grave failure
and I would appreciate knowing more about your setup/configuration if
possible.
> I am beginning to think that zeroconf should, in the ifupdown world,
> either be a distinct configuration method or an option for the dhcp
> method.
My inital thought was to implement zeroconf (IPv4LL) as a fall-back for
the various DHCP clients. However section 2.11 notes that modifyin the
DHCP state machine to incorporate zeroconf "reduces the reliability of
the DHCP service".
Since the specification does note that IPv4LL and IPv4 global scope will
be in use simultaneously and has merely a 'SHOULD NOT' prohibition (i.e.
don't do it if you don't know what you are doing) I believe configuring
zeroconf address via the if-up.d approach is correct.
Cheers,
Anand
PS: Thank you for your bug reports, they've been most helpful and have
forced me to re-read the specification to clarify my justification.
--
`When any government, or any church for that matter, undertakes to say to
its subjects, "This you may not read, this you must not see, this you are
forbidden to know," the end result is tyranny and oppression no matter how
holy the motives' -- Robert A Heinlein, "If this goes on --"
signature.asc
Description: Digital signature
