Re: [Ddclient-support] Support for IPv6 (patch)

[Eduardo T. <etr...@gm...>]

>> I wrote a little patch for 3.8.1 to support IPv6 on the two services
>> I know that can handle it.
> 
> I'm not sure i understand.
> 
> DynDNS for IPv6 sounds like a terrible misconcept, a contradiction
> all in itself.

I'm using a teredo client (transition technology) so I get an dynamic
IPv6 over an IPv4 tunnel, I'd like to be able to access it.  I need a
name for that because doing:

ssh user@[2001:0000:4136:e378:8000:63bf:3fff:fdd2]

is not practical or, at least not so simple as:

ssh us...@te...

when you take into account that the host is miles aways from me, behind
a NAT router that changes its IPv4 address periodically.  I'm actually
(and happily) using a name for dynamic IPv6 right now.  So much for the
misconcept or contradiction.  IPv4 just would not work, or I would have
to setup a VPN for incoming connections ( and believe me it's a lot
easier to do apt-get install miredo, and that's it).

> I thought the whole point of IPv6 is to have enough addresses
> such that everybody can have their own, proper addresses?

We found the misconcept and inner contradiction.  There it is.  Routing
and latency should ring an alarm with that idea of yours, not to mention
routing size tables.  If you fly from the US to France, even if possible
(as with IPv4) I doubt you would want to keep "your own" IPv6 address
from the other side of the ocean for your mobile or your notebook.  And
no, nobody would care to change transatlantic routes just for a user.
You would most likely want to pick up the hotel room or the university
IPv6 address, or the airport wireless one and then, you host will be
more easily found by using a name (dyndns to the rescue!). Of course
there are extensions for handling mobile, multihoming and staff like
that in IPv6 (also for IPv4) for example [1][2] but they are overkill
for most individuals.

Everybody is going to be addressable in IPv6, but nobody garantees you
will always have the same address wherever you go!!  That's at least
naïve if you consider the impact that would have for moving users on the
size of the backbone routing tables.

So, may I suggest that the misconception lies on your side?  If we had
enough IPv4 you would consider dynamic DNS "a terribly misconcept"?  The
key concept here is not *address* but *dynamic*.

>> "ip -6 addr list" instead of "ifconfig" because I find it easier
>> to parse and it doesn't need root privileges (Debian box).
> 
> Don't do that.  The /sbin/ip command is utterly non-portable
> and purely Linux-only, so it should not be used for anything
> outside purely Linux-only contexts.

Ok, if you think this can make it into mainline I can change that.

> Besides, using ifconfig(8) to query an IP address does not
> require root privileges.

No, it doesn't, you're right.  It's simply that Debian doesn't have
/sbin int its path for non root users.  You have to add it or call
/sbin/ifconfig directly but you do not need root privileges.

> Even if the basic concept would make any sense, i don't see
> any patch anywhere.

The list didn't just let it through, please care to read the next
message on the same subject to find the link to the actual patch.  I
copy it here[3] just in case.

About the basic concept, you might want to reconsider it.  I'm not
proposing it, I'm using it already from dyndns.com and I've used it from
afraid.org.  There will be dynamic IPv6, and not only during transition
(which will last long enough).  Happily somebody is already letting us
avoid things like 2001:0000:4136:e378:8000:63bf:3fff:fdd2 and using
names instead.  Teredo and 6to4 (on dynamic IPv4) are not rare, I just
tried to add support to ddclient because that's the client I use and
love, I tried to take the least invasive route (just a keyword).

It's your call to ignore the patch, rewrite it or do as you wish,
dynamic dns for IPv6 is out there, it's used and eventually, if it
proves useful it will be soported, if not by upstream as a patch in
Debian, with ifconfig not iproute2, I get it.  In any case, I suggest
you revisit the whole IPv6 concept.

Friendly, Eduardo.

[1] Mobility Support in IPv6 http://tools.ietf.org/html/rfc6275
[2] Provider-independent address space
https://www.ripe.net/ripe/docs/ripe-510#2e
[3]
https://sourceforge.net/tracker/?func=detail&aid=3407066&group_id=116817&atid=676131