What is the proper entry for /etc/hosts 127.0.0.1?

[David C. Rankin]

Guys,

What is the proper entry for 127.0.0.1 in /etc/hosts? I have seen discussion
that 127.0.0.2 is deprecated so that leaves a problem with dnsdomainname if you
remove the 127.0.0.2 entry and leave the 127.0.0.1 entry as:

127.0.0.1 localhost.localdomain localhost yourRealHostname

In this situation, dnsdomainname resolves to localdomain:

15:39 phoenix:~> dnsdomainname
localdomain

Setting the entry as:

127.0.0.1 localhost.youReadDomainName.com localhost yourRealHostname

fixes the return of dnsdomainname, but does not having localdomain cause
problems as well? What say the gurus?
--
David C. Rankin, J.D.,P.E.
Rankin Law Firm, PLLC
510 Ochiltree Street
Nacogdoches, Texas 75961
Telephone: [phone]
Facsimile: [phone]
www.rankinlawfirm.com

[Per Jessen]

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts?

My systems all have:

127.0.0.1 localhost

Post by David C. Rankin
I have seen discussion that 127.0.0.2 is deprecated

Like John says, 127.0.0.0/8 are usually assigned to the loopback
interface, and are usually = localhost.

[John Andersen]

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts? I have seen discussion
that 127.0.0.2 is deprecated so that leaves a problem with dnsdomainname if you
127.0.0.1 localhost.localdomain localhost yourRealHostname
15:39 phoenix:~> dnsdomainname
localdomain
127.0.0.1 localhost.youReadDomainName.com localhost yourRealHostname
fixes the return of dnsdomainname, but does not having localdomain cause
problems as well? What say the gurus?

I thought averything 127.x.x.x pointed to localhost.
Did I miss a memo?

[Per Jessen]

Post by Per Jessen

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts?

127.0.0.1 localhost

Post by David C. Rankin
I have seen discussion that 127.0.0.2 is deprecated

Like John says, 127.0.0.0/8 are usually assigned to the loopback
interface, and are usually = localhost.

IIRC, OpenSUSE assigns 127.0.0.1 to local host and 127.0.0.2 to the
computer name.

Isn't there an option for that during installation? I seem to remember
some network option I always deselect.

[James Knott]

Post by Per Jessen

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts?

127.0.0.1 localhost

Post by David C. Rankin
I have seen discussion that 127.0.0.2 is deprecated

Like John says, 127.0.0.0/8 are usually assigned to the loopback
interface, and are usually = localhost.

IIRC, OpenSUSE assigns 127.0.0.1 to local host and 127.0.0.2 to the
computer name.

It's much simpler on IPv6, where ::1 works for both. ;-)

[James Knott]

Post by Per Jessen

IIRC, OpenSUSE assigns 127.0.0.1 to local host and 127.0.0.2 to the

computer name.

Isn't there an option for that during installation? I seem to remember
some network option I always deselect.

Yes, it asks if you want the computer name to be added to the hosts
file. I generally allow that, as it doesn't hurt anything and may be
useful on servers. So, on my home network, I'm running dmsmasq on my
firewall, which has all the computers and devices listed in /etc/hosts
(with the IPv6 address listed first). Then all the computers have
localhost & computer name in their own hosts file. The only exception
is my notebook, which has all the devices listed in it's hosts file, as
I often use it to connect to stuff at home via IPv6 or VPN.

[David C. Rankin]

Post by Per Jessen

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts?

127.0.0.1 localhost

Post by David C. Rankin
I have seen discussion that 127.0.0.2 is deprecated

Like John says, 127.0.0.0/8 are usually assigned to the loopback
interface, and are usually = localhost.

Well,

The documentation shows (man hosts):

IP_address canonical_hostname [aliases...]

showing you can have any number of aliases for a given IP_address
canonical_hostname pair. So I have always included the hostname as an alias
after localhost:

127.0.0.1 localhost phoenix

which provides resolution for localhost as well as your hostname on loopback. I
have run across several web apps that want to be able to resolve the loopback
address with the hostname. The 127.0.0.2 SuSE has used is a hack and there is
talk about doing away with it completely.

The real problem is with dnsdomainname (or hostname -d) which will not resolve
to anything unless you include the full canonical_hostname in /etc/hosts:

127.0.0.1 providence.rlfpllc.com localhost providence

This works fine for dnsdomainname (or hostname -d). Also, since it provides the
definition for localhost on loopback, it does what it is supposed to as far as I
can tell. I can't see any downside to this approach, but I'm skeptical enough to
ask and see if anybody else knows of any gotchas or side-effects from doing this?
--
David C. Rankin, J.D.,P.E.
Rankin Law Firm, PLLC
510 Ochiltree Street
Nacogdoches, Texas 75961
Telephone: [phone]
Facsimile: [phone]
www.rankinlawfirm.com

[James Knott]

I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).

That would break a lot of existing computers and devices. The proper
way out of the address shortage is IPv6. I have my own IPv6 /56 subnet
with an address space about 1 trillion times the entire IPv4 address
space. If only /48 subnets were handed out, as some propose, there's be
enough addresses for about 35 trillion of them, each containing a
trillion, trillion addresses and this is using only the 1/8 of the
entire IPv6 address range that's currently allocated to unicast addresses.

[Philipp Thomas]

I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).

Restricting the range of localhost would be a dumb move as 127.0.0.0/8 has
been in effect for far too long and is embedded in too many devices. Face
it, IPv6 is the only viable way to go.

Philipp

[James Knott]

there's be enough addresses for about 35 trillion of them

Correction, that should be 32 trillion.

[James Knott]

Post by Philipp Thomas

I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).

Restricting the range of localhost would be a dumb move as 127.0.0.0/8 has
been in effect for far too long and is embedded in too many devices. Face
it, IPv6 is the only viable way to go.

Even if it was only 127.0.0.0/8, it would only delay the inevitable.
IPv4 simply cannot meet the needs of today, let alone the future. The
more people try to dream up schemes to extend IPv4, the worse the
problem will get. Move to IPv6. I have!

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Even if it was only 127.0.0.0/8, it would only delay the inevitable. IPv4
simply cannot meet the needs of today, let alone the future. The more people
try to dream up schemes to extend IPv4, the worse the problem will get. Move
to IPv6. I have!

It is not our decision to make. It is our providers (the ISPs) who must
make that move.

And not one provider, but all, at the same time.

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0j7bEACgkQtTMYHG2NR9VatwCeI2zeVnw+SxzSPkkFojvEwW+5
2kYAn2QBj15INQudnvr4n8zp5IbBqGsx
=5fjy
-----END PGP SIGNATURE-----

[Hans Witvliet]

Post by Carlos E. R.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Even if it was only 127.0.0.0/8, it would only delay the inevitable. IPv4
simply cannot meet the needs of today, let alone the future. The more people
try to dream up schemes to extend IPv4, the worse the problem will get. Move
to IPv6. I have!

It is not our decision to make. It is our providers (the ISPs) who must
make that move.

No, it is yours.
Don't use your ISP-IPv6-reluctance for any delay on your side.

Around the net you can find reports that failure to provide IPv6 will be
a reason for ISP's to go bust. Are you willing to wait for that to
happen? And then be forced to move to another ISP?

In the mean time, use a tunnelbroker, while you can not get native-IPv6.
I use HE (tunnelbroker.net, but there are others) in the mean time. My
provider has started to give V6-addresses for those who want it, but i
seriously think of sticking to my independant-tunnel-provider, instead
of going native.
My /48 is now independant of my isp,and i can switch to another isp
without any re-numbering, Just adjusting my tunnel setup.

hw

[Wilfred van Velzen]

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get native-IPv6.
I use HE (tunnelbroker.net, but there are others) in the mean time. My
provider has started to give V6-addresses for those who want it, but i
seriously think of sticking to my independant-tunnel-provider, instead
of going native.
My /48 is now independant of my isp,and i can switch to another isp
without any re-numbering, Just adjusting my tunnel setup.

http://www.fix6.net/archives/2010/12/06/the-trouble-with-6to4-2/
http://www.potaroo.net/ispcol/2010-12/6to4fail.html

I think you should allways choose native, if you have the choice.

And your also dependent on the bottleneck at the tunnel provider. Will they invest enough and on time in new equipment, when they become more widely used?
--
Met vriendelijke groet / Best regards,
Wilfred van Velzen
--
SERCOM Regeltechniek B.V.
Heereweg 9
2161 AB Lisse
Nederland
[phone] (voice)
[phone] (fax)

<http://www.sercom.nl/>


Op al onze offertes, op alle opdrachten aan ons en op alle met ons gesloten
overeenkomsten zijn toepasselijk de METAALUNIEVOORWAARDEN, gedeponeerd ter
Griffie van de Rechtbank te Rotterdam, zoals deze luiden volgens de
laatstelijk aldaar neergelegde tekst. De leveringsvoorwaarden worden u op
verzoek toegezonden.
---

[Dave Howorth]

Post by Hans Witvliet
No, it is yours.
Don't use your ISP-IPv6-reluctance for any delay on your side.
Around the net you can find reports that failure to provide IPv6 will be
a reason for ISP's to go bust. Are you willing to wait for that to
happen? And then be forced to move to another ISP?
In the mean time, use a tunnelbroker ... [snip] ...
My /48 is now independant of my isp,and i can switch to another isp
without any re-numbering, Just adjusting my tunnel setup.

I'm pretty ignorant of the issues, so could you or somebody else explain
how this helps? (perhaps a link to a RTFM/FAQ?)

Surely, if your ISP goes broke then you will still have to change your
mail, DNS, ADSL login etc credentials anyway (or else the ISP must have
been taken over seamlessly) so I'm not clear how already having an IPV6
connection saves much work.

Cheers, Dave

[Per Jessen]

Post by Dave Howorth

Post by Hans Witvliet
No, it is yours.
Don't use your ISP-IPv6-reluctance for any delay on your side.
Around the net you can find reports that failure to provide IPv6 will
be a reason for ISP's to go bust. Are you willing to wait for that to
happen? And then be forced to move to another ISP?
In the mean time, use a tunnelbroker ... [snip] ...
My /48 is now independant of my isp,and i can switch to another isp
without any re-numbering, Just adjusting my tunnel setup.

I'm pretty ignorant of the issues, so could you or somebody else
explain how this helps? (perhaps a link to a RTFM/FAQ?)
Surely, if your ISP goes broke then you will still have to change your
mail, DNS, ADSL login etc credentials anyway (or else the ISP must
have been taken over seamlessly) so I'm not clear how already having
an IPV6 connection saves much work.

If you're running services (mail, web, whathaveyou) on your IPv6 range,
a change of provider will mean changing IP-addresses, just like today.
If you're not running services, it doesn't matter much - just like
today.

[Dave Howorth]

Post by Per Jessen

Post by Dave Howorth
Surely, if your ISP goes broke then you will still have to change your
mail, DNS, ADSL login etc credentials anyway (or else the ISP must
have been taken over seamlessly) so I'm not clear how already having
an IPV6 connection saves much work.

If you're running services (mail, web, whathaveyou) on your IPv6 range,
a change of provider will mean changing IP-addresses, just like today.
If you're not running services, it doesn't matter much - just like
today.

Right, but using a tunneled IPv6 now doesn't seem to change the overall
amount of work. Roughly speaking, either:

(a) my ISP is stupid and goes broke or is so badly organized that I have
to change to and set up another, or
(b) my ISP is wise and I'm automatically migrated to IPv6 (or he sends
me clear instructions on what I need to do)

But I don't see how me doing anything now separately from my ISP reduces
my overall workload. It feels likely that it will actually increase the
work.

( Not trying to start a fight, just to understand if Hans' suggestion
has legs. :)

Cheers, Dave

[Per Jessen]

Post by Dave Howorth
Right, but using a tunneled IPv6 now doesn't seem to change the
(a) my ISP is stupid and goes broke or is so badly organized that I
have to change to and set up another, or
(b) my ISP is wise and I'm automatically migrated to IPv6 (or he sends
me clear instructions on what I need to do)
But I don't see how me doing anything now separately from my ISP
reduces my overall workload. It feels likely that it will actually
increase the work.

Yeah, I think that's about it. Today, for the regular end-user, IPv6 is
just something to play with if one is so inclined.

[Hans Witvliet]

Post by Dave Howorth

Post by Per Jessen

Post by Dave Howorth
Surely, if your ISP goes broke then you will still have to change your
mail, DNS, ADSL login etc credentials anyway (or else the ISP must
have been taken over seamlessly) so I'm not clear how already having
an IPV6 connection saves much work.

If you're running services (mail, web, whathaveyou) on your IPv6 range,
a change of provider will mean changing IP-addresses, just like today.
If you're not running services, it doesn't matter much - just like
today.

Right, but using a tunneled IPv6 now doesn't seem to change the overall
(a) my ISP is stupid and goes broke or is so badly organized that I have
to change to and set up another, or
(b) my ISP is wise and I'm automatically migrated to IPv6 (or he sends
me clear instructions on what I need to do)
But I don't see how me doing anything now separately from my ISP reduces
my overall workload. It feels likely that it will actually increase the
work.
( Not trying to start a fight, just to understand if Hans' suggestion
has legs. :)
Cheers, Dave

Just as Per wrote,
I assume you have you mail and other services on you own domain-name,
and not on a sub-domain of your provider.
So mail/web/other-traffic goes to yourdomain.org

Post by Dave Howorth
From your current-isp you get an ip-address and on that adres you

terminate your IPv6-tunnel. Along with your tunnel you get an /64
network (ad if you want an extra /48 network)
Define all of your services of the addresses of that /48 network.

In case you switch to another provider, only the ipv4-address of your
tunnel changes, nothing else.

##
Actually, this is more related to managing your own domainname, but
since i started to use v6, i rather use names instead of addresses..
But even when you just use plain numbers, it is just adjusting your
tun-setup script with the ip-adres of your new provider, as the
addresses of your /48 remains the same.

hw

[Dave Howorth]

Post by Hans Witvliet
Just as Per wrote,
I assume you have you mail and other services on you own domain-name,
and not on a sub-domain of your provider.
So mail/web/other-traffic goes to yourdomain.org

I think this is the key. My mail goes via the mail server of my ISP (in
fact I think it may have to, to prevent open relays, though I'm not
sure). My DNS comes from my ISP. So I'm just a regular user as I suspect
Carlos is in this regard, though I seem to have more choice of ISP!

(I do have my own domain, but it is something completely separate to
which I just have net connections)

So I think Per's later assessment is about right. For people in my
position at present the only reason to use IPv6 is if I view it as
educational or 'fun'. Which in some circumstances I might, but I've got
other things to keep me busy at present.

Thanks for the info anyway,
Dave

[Hans Witvliet]

Post by Wilfred van Velzen

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get native-IPv6.
I use HE (tunnelbroker.net, but there are others) in the mean time. My
provider has started to give V6-addresses for those who want it, but i
seriously think of sticking to my independant-tunnel-provider, instead
of going native.
My /48 is now independant of my isp,and i can switch to another isp
without any re-numbering, Just adjusting my tunnel setup.

http://www.fix6.net/archives/2010/12/06/the-trouble-with-6to4-2/
http://www.potaroo.net/ispcol/2010-12/6to4fail.html
I think you should allways choose native, if you have the choice.
And your also dependent on the bottleneck at the tunnel provider. Will they invest enough and on time in new equipment, when they become more widely used?
--

Well, instinctively i would agree: keep it as simpel as possible.

But performance is good: sometimes my latency is slightly (1 ms) higher.
And horror stories like above (failure-rate) are completely unknown.

Since HE has a end-point on the ams-ix, performance is that good, that i
even recommended it at work. Furthermore, HE has 25 tunnel-endpoints at
several major internet-router-sites (london, frankfurt, paris,
singapore, hongkong), so you can avoid the single point of failure.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Hans Witvliet

Post by Carlos E. R.
It is not our decision to make. It is our providers (the ISPs) who must
make that move.

No, it is yours.
Don't use your ISP-IPv6-reluctance for any delay on your side.

I can't move unles:

My ISP changes to IPv6
The rest of the world changes

If not, I will not be able to connect. What advantage do I get, if all
sites I connect to are on IPv4?

Post by Hans Witvliet
Around the net you can find reports that failure to provide IPv6 will be
a reason for ISP's to go bust. Are you willing to wait for that to
happen? And then be forced to move to another ISP?

Absolutely yes.

For one thing, they will not go bust, they'll move. They are too big, and
the state has interests on them. They fail, then no telephone in the
country.

For another, there are no other ISPs. None here offers IPv6, and all of
them except one uses the infraestructure of the main ISP, for starters.

So, I wait for them to move and force me to change - meaning that they
will have to pay for the router.

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get native-IPv6.

I don't see any advantage.

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0kT48ACgkQtTMYHG2NR9UosQCdHRwguCmQTsP1TIhrMslCCPmN
vLwAn0MYwLE6JGU2J8myy52a6+tHKwg8
=/xpD
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.

Post by Hans Witvliet

Post by Carlos E. R.
It is not our decision to make. It is our providers (the ISPs) who
must make that move.

No, it is yours.
Don't use your ISP-IPv6-reluctance for any delay on your side.

My ISP changes to IPv6
The rest of the world changes

Not quite correct - like Hans suggests, you could use a tunnelling
setup, or use 192.88.99.1 as a 6to4 relay. The rest of the world is
already changing, even if only slowly.

Post by Carlos E. R.
If not, I will not be able to connect. What advantage do I get, if all
sites I connect to are on IPv4?

None (except the experience in setting it up).

Post by Carlos E. R.

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get
native-IPv6.

I don't see any advantage.

There aren't really any.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Per Jessen

Post by Carlos E. R.
My ISP changes to IPv6
The rest of the world changes

Not quite correct - like Hans suggests, you could use a tunnelling
setup, or use 192.88.99.1 as a 6to4 relay. The rest of the world is
already changing, even if only slowly.

I could move, but what do /I/ gain?
Nothing that I can see.

Post by Per Jessen

Post by Carlos E. R.

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get
native-IPv6.

I don't see any advantage.

There aren't really any.

That's the point.

Doubtless, we we'll have to move. But why should I go first?

I know nothing about IPv6, not even enough to use it in my own little home
network. All the sites I connect to are on v4. I would have to invest on
hardware, and when my ISP does the inevitable and changes, then perhaps
I'll have to invest again.

So no. They change first and carry us all. Like sheep in a flock. :-p

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0kVWcACgkQtTMYHG2NR9XiPgCfTos0QufENiBN4DNwj+5988hX
/t8AnRpxvFO4xuEdT4aoDVX5GDepP8+v
=SCv2
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.

Post by Hans Witvliet
In the mean time, use a tunnelbroker, while you can not get
native-IPv6.

I don't see any advantage.

There aren't really any.

That's the point.
Doubtless, we we'll have to move. But why should I go first?
I know nothing about IPv6, not even enough to use it in my own little
home network. All the sites I connect to are on v4. I would have to
invest on hardware, and when my ISP does the inevitable and changes,
then perhaps I'll have to invest again.

Well, no, as Hans has described, you could use a 6to4 tunnel - they're
available for free and requires no addtional hardware. No, it brings
you no advantages, it's only for you to play with.

[James Knott]

I didn't realize we're up to IPv8 already. What happened to 6 & 7? ;-)

[James Knott]

Post by Carlos E. R.
I could move, but what do /I/ gain?
Nothing that I can see.

You gain access to all the available web sites, not just those on IPv4.
You get direct access to all your computers, without NAT getting in the
way. In short, you get to use the internet in the manner it was
intended to be used.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Carlos E. R.
I could move, but what do /I/ gain?
Nothing that I can see.

You gain access to all the available web sites, not just those on IPv4. You
get direct access to all your computers, without NAT getting in the way. In
short, you get to use the internet in the manner it was intended to be used.

I can not even reach my local printer on IPv6...

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lQqsACgkQtTMYHG2NR9U5iQCfcw91epFCCvNVihYx5daCaXaZ
xesAnRKcxcHS46GexLrjcqOAi290isIO
=OmcD
-----END PGP SIGNATURE-----

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by James Knott
I didn't realize we're up to IPv8 already. What happened to 6 & 7? ;-)

Yeah, I figured v6 was old already, with all those galaxies to reach.
Not enough numbers.

:-p

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lQvsACgkQtTMYHG2NR9U+8ACfSdXua4tO9xq4tI5pGYvZpbEu
bxQAoIbjMSPaLMGTBlPBbC5WaxjhRM+m
=dlcZ
-----END PGP SIGNATURE-----

[James Knott]

Post by Carlos E. R.
My ISP changes to IPv6
The rest of the world changes
If not, I will not be able to connect. What advantage do I get, if all
sites I connect to are on IPv4?

Yes you can. Operating systems such as Linux, Windows, Mac etc. can run
dual stack, which means you can access both IPv4 and IPv6 sites.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Carlos E. R.
If not, I will not be able to connect. What advantage do I get, if all
sites I connect to are on IPv4?

Yes you can. Operating systems such as Linux, Windows, Mac etc. can run dual
stack, which means you can access both IPv4 and IPv6 sites.

I doubt it. I can not reach my own local printer, which I know supports
v6. If that doesn't work, even less the outside.

Access my local machines without NAT? I prefer not to. I have more to lose
than gain :-P

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lQ3IACgkQtTMYHG2NR9UdfACeJwkIkfPIJ9BHfQqaQOS5h8QT
LVQAnj/U04oOY6ETzfCOJCwNqnFE7bZ+
=3QpS
-----END PGP SIGNATURE-----

[James Knott]

Post by Carlos E. R.
I doubt it. I can not reach my own local printer, which I know
supports v6. If that doesn't work, even less the outside.

I'm not sure why you're having the problem. I'll have to look into it
when I have more time. However, those link local IPv6 addresses are
awkward to use, compared to the addresses you'd normally use.

Post by Carlos E. R.
Access my local machines without NAT? I prefer not to. I have more to
lose than gain :-P

You lose nothing that a properly set up wirewall will give you.

[Ken Schneider - openSUSE]

Post by James Knott

Post by Carlos E. R.
I doubt it. I can not reach my own local printer, which I know
supports v6. If that doesn't work, even less the outside.

I'm not sure why you're having the problem. I'll have to look into it
when I have more time. However, those link local IPv6 addresses are
awkward to use, compared to the addresses you'd normally use.

Post by Carlos E. R.
Access my local machines without NAT? I prefer not to. I have more to
lose than gain :-P

You lose nothing that a properly set up wirewall will give you.

I, like a lot of other people, do not run any local services so although
I have nothing to lose I also have absolutely nothing to gain.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Ken Schneider - openSUSE

Post by James Knott

Post by Carlos E. R.
Access my local machines without NAT? I prefer not to. I have more to
lose than gain :-P

You lose nothing that a properly set up wirewall will give you.

I, like a lot of other people, do not run any local services so although
I have nothing to lose I also have absolutely nothing to gain.

That is so.

VoIP would be an exception, but then, on the against side, is p2p, it is
preferable to be anonymous ;-)

About that "properly set up wirewall", I don't have one. I have what my ISP
gave me, which is only capable of v4 things, and limited at that. And if I
go to the local shops, I'm sure I would only find similar adsl routers,
none capable of ipv6 - which would be useless because the providers do not
offer it.

I would need a box capable of tunneling over IPv4, and I wouldn't have the
foggiest idea what to buy. And it would be expensive, so I'll better wait
till the ISPs do their move.

I have no hurry :-)

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAk0l/CYACgkQtTMYHG2NR9XIyACggYka30ueH7JS4b5y58ppaHMn
ViMAn2qOiZxwq86Qb941IhY9oKpmBvko
=5ZYb
-----END PGP SIGNATURE-----

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Content-ID: <***@Telcontar.valinor>


Resending, my ISP is failing again :-/

<***@opensuse.org>: host smtp.telefonica.net[213.4.149.228] said: 552
5.2.0 sKms1f00J1LWEzc1jKmtS6 internal error (in reply to end of DATA
command)

You trust the one-size-fits-all firewall that your ISP has provided?

Not really :-)

Why not run your own, e.g. the openSUSE firewall?

I do, it is on, but it only protects the computer. There are devices
connected to the router without their own firewall: the printer, the
digital TV recorder...

Yes, I know I can get a computer with linux, two ethernet ports, and a
switch, to really protect the local network. But that is more expense and a
good deal of hassle. And more electricity.

Post by Carlos E. R.
I would need a box capable of tunneling over IPv4, and I wouldn't have
the foggiest idea what to buy.

Your own Linux box is all you need.

Yes, for services from the computer. If I'd want (which I don't) to put a
lone VoIp phone over IPv6, I would need something else. That is, if I want
to get some of the advantages of IPv6, I would need something else.

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0mI9kACgkQtTMYHG2NR9WqHQCfYCzQZ1j2yGpjewlaMgkYA3wl
C/kAn2Lo+Dv/wMVz7V7+FJiQ6YrHv+B6
=1mj+
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.
About that "properly set up wirewall", I don't have one. I have what
my ISP gave me, which is only capable of v4 things, and limited at
that. And if I go to the local shops, I'm sure I would only find
similar adsl routers, none capable of ipv6 - which would be useless
because the providers do not offer it.

You trust the one-size-fits-all firewall that your ISP has provided?
Why not run your own, e.g. the openSUSE firewall?

Post by Carlos E. R.
I would need a box capable of tunneling over IPv4, and I wouldn't have
the foggiest idea what to buy.

Your own Linux box is all you need.

[jsa]

On Thu, 6 Jan 2011 21:19:36 +0100 (CET), "Carlos E. R."

Post by Carlos E. R.
I do, it is on, but it only protects the computer. There are devices
connected to the router without their own firewall: the printer, the
digital TV recorder...

Get a router that supports alternative firmwares like DDWRT and run
the lfirewall on it.
Philipp

But Getting a new router was an expense Carlos was looking to avoid.

To me, the easiest thing seems to be to do this with a Linux machine
as the only thing attached to the ISP.

Its not that much more expensive than running a router, presuming you
HAVE an old machine laying around (which I suspect most of us do).
Running Headless, the power consumption is miniscule, its just as easy
to configure as a router, and easier to keep up to date.

That way during the next year or two of flux in the ipv4/ipv6 arena, you
will be able to roll with the punches delivered by your ISP, adopt ipv6
well ahead of them if you wish, and use the linux box to do your tunneling.

I do this now, and run linux as my gateway/router. My WiFi is relagated
to merely an Access Point to my LAN. It doesn't even do dhcp.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Post by jsa

Get a router that supports alternative firmwares like DDWRT and run
the lfirewall on it.

But Getting a new router was an expense Carlos was looking to avoid.

Correct.

Post by jsa
To me, the easiest thing seems to be to do this with a Linux machine
as the only thing attached to the ISP.

No, the easiest is doing nothing :-P

A PC is not enough. For one, it doesn't have an ADSL port, so I still need the router from the
provider. Then, the old PC I have has only one eth port, and for a router you need two - being an
old computer, it might be impossible to get that card. And having very little memory (the one I have
has only 32 Megs) it can run very few distros, if any at all. Then, on the home side ethernet port,
I would have to plug a switch to connect my local machines. I would also need a wifi port to feed
the laptop.

That's four machines to replace the existing router.

A lot of space (I already have a 2 meter high iron rack) and electricity (and maintenance effort)
for something I do not need at all to do.


If I were to do it, I would prefer some dedicated little box, a real router, preferable with open
source code, as Philip says - but as I have no need for that, I'll do nothing till the current setup
breaks down (at which moment I'll buy whatever I can get locally in a hurry), or till the ISP pushes
upgrades down to the users - which in any case I would be forced to do when they do decide to change
to IPv6.

If someone has a suggestion of a real good and cheap router, with adsl port, wifi, generic code,
support for IPv4, V6, and whatever is needed now and for the future (wow!), then speak up! :-)

Not that I would buy that right now, but I would take a note for the future.

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" GM (Minas Tirith))
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iF4EAREIAAYFAk0mazwACgkQja8UbcUWM1yCPQD+ImzAadCYocDEOm5jEGaLuGkU
pgq3Lp0kftQJE0i3drMBAIKtjWSUbltYiMct2KV7wuIdxwmDm+3SSWHHYZa27NMw
=EqY4
-----END PGP SIGNATURE-----

[James Knott]

Post by Carlos E. R.
About that "properly set up wirewall", I don't have one. I have what my ISP
gave me, which is only capable of v4 things, and limited at that. And if I
go to the local shops, I'm sure I would only find similar adsl routers,
none capable of ipv6 - which would be useless because the providers do not
offer it.
I would need a box capable of tunneling over IPv4, and I wouldn't have the
foggiest idea what to buy. And it would be expensive, so I'll better wait
till the ISPs do their move.

You can get boxes from Linksys & D-link, among others that should be
easy to set up and not too expensive.

[Per Jessen]

Post by jsa

On Thu, 6 Jan 2011 21:19:36 +0100 (CET), "Carlos E. R."

Post by Carlos E. R.
I do, it is on, but it only protects the computer. There are devices
connected to the router without their own firewall: the printer, the
digital TV recorder...

Get a router that supports alternative firmwares like DDWRT and run
the lfirewall on it.
Philipp

But Getting a new router was an expense Carlos was looking to avoid.
To me, the easiest thing seems to be to do this with a Linux machine
as the only thing attached to the ISP.

Yep. Any old PC will do.

[Per Jessen]

Post by Carlos E. R.
A PC is not enough. For one, it doesn't have an ADSL port, so I still
need the router from the provider. Then, the old PC I have has only
one eth port, and for a router you need two - being an old computer,
it might be impossible to get that card.

If you need ISA, I'll send you a couple of 3c590 or 3c515 if you pay the
postage. If PCI works, you can have a D-Link <somewthing> with RA8381.

Post by Carlos E. R.
And having very little memory (the one I have has only 32 Megs) it can
run very few distros, if any at all.

I ran my internet gateway on a 486DX2 with 24Mb for years. I think it
was SuSE Linux 7.1 - with 11.3 the issue is the installation, not the
operation. You might have to add a bit of memory during installation.
Anyway, I can pop some old memory in the package when I send the
3c590 - in fact, you have the whole box, but it'll be a bit expensiveto
ship :-)

Today I would probably buy some embedded thingie, more probably one of
these http://www.pcengines.ch/ and then run the xDSL router in
bridging mode. Setting it up would be more work, but I would want the
flexibility.

Post by Carlos E. R.
If someone has a suggestion of a real good and cheap router, with adsl
port, wifi, generic code, support for IPv4, V6, and whatever is needed
now and for the future (wow!), then speak up! :-)

I don't know of one that will satisfy all of those requirements. Quite a
few come close though - Zyxel has very affordable boxes, but only one
that does IPv6 and it's not using openWRT. Linksys WRT54GL is good
option, but it doesn't have ADSL, so you'd have to keep your existing
modem.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Post by Per Jessen

Post by Carlos E. R.
A PC is not enough. For one, it doesn't have an ADSL port, so I still
need the router from the provider. Then, the old PC I have has only
one eth port, and for a router you need two - being an old computer,
it might be impossible to get that card.

If you need ISA, I'll send you a couple of 3c590 or 3c515 if you pay the
postage. If PCI works, you can have a D-Link <somewthing> with RA8381.

Well, thanks, but better not. I feel lazy in this matter, it is not something I really need so I'll
probably do nothing in a long time O:-)

I was only describing the difficulties, not calling for help - not yet, anyway. I could probably get
a card second hand quite cheap.

Post by Per Jessen
I ran my internet gateway on a 486DX2 with 24Mb for years. I think it
was SuSE Linux 7.1 - with 11.3 the issue is the installation, not the
operation.

The machine has 7,3 currently.

Post by Per Jessen
Today I would probably buy some embedded thingie, more probably one of
these http://www.pcengines.ch/ and then run the xDSL router in
bridging mode. Setting it up would be more work, but I would want the
flexibility.

Yes, I had thought of that, too. That idea tickles me a bit :-)

Post by Per Jessen

Post by Carlos E. R.
If someone has a suggestion of a real good and cheap router, with adsl
port, wifi, generic code, support for IPv4, V6, and whatever is needed
now and for the future (wow!), then speak up! :-)

I don't know of one that will satisfy all of those requirements. Quite a
few come close though - Zyxel has very affordable boxes, but only one
that does IPv6 and it's not using openWRT. Linksys WRT54GL is good
option, but it doesn't have ADSL, so you'd have to keep your existing
modem.

I'll have a (lazy) look to see what I can find, thanks ;-)

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" GM (Minas Tirith))
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iF4EAREIAAYFAk0m5rUACgkQja8UbcUWM1yUDwEAmJbL9nA5cdCDC3K5A1iwB32F
Rnmfa9TTXNMXGAD+M6QBAJmvZuV9Ymi1QFXKGxA73K6mYV/57zQiqfW4Vk2c0vpI
=y03I
-----END PGP SIGNATURE-----

[James Knott]

Post by jsa
I do this now, and run linux as my gateway/router. My WiFi is relagated
to merely an Access Point to my LAN. It doesn't even do dhcp.

That's exactly what I do here. My firewall is an old Compaq* computer,
running OpenSUSE 11.3. It provides firewall, router, DHCP, DNS, VPN end
point etc. Like you, my WiFi is a router used in bridging mode.

* I wonder if a Gateway would work better? ;-)

[James Knott]

Post by Ken Schneider - openSUSE
I also have absolutely nothing to gain.

At the moment. However, there are other things in the works that make
accessing your home network desirable. One simple example is home
control & security. You could control your thermostat etc. remotely.
People currently use those X11 modules for that. You can also check
status & alarms or even security cameras remotely. Many, many more
things are possible.

[Ken Schneider - openSUSE]

Post by James Knott

Post by Ken Schneider - openSUSE
I also have absolutely nothing to gain.

At the moment. However, there are other things in the works that make
accessing your home network desirable. One simple example is home
control & security. You could control your thermostat etc. remotely.
People currently use those X11 modules for that. You can also check
status & alarms or even security cameras remotely. Many, many more
things are possible.

As I said _I_ have absolutely nothing to gain.

[James Knott]

Post by Carlos E. R.
It is not our decision to make. It is our providers (the ISPs) who
must make that move.

No, you can make the move right now, the same way I did. You use a
service called a "tunnel broker". With a tunnel broker, you set up your
computer or firewall to use 6in4 tunnelling. This adds a 20 byte IPv4
header to the IPv6 packet, for transport over the existing IPv4
network. The tunnel broker then removes that header and puts your IPv6
packet on the IPv6 internet. You can do this with Linux, Windows, Mac
and other. There are also many routers that support this. Cisco
routers certainly do, as do certain D-link models. I believe the Apple
Airports do as well. Also, given Cisco's support for IPv6, I wouldn't
be surprised to see some Linksys models with it to. It can also be used
with the various WRT models that can be reflashed with (IIRC) OpenWRT.
My home network, including DNS is set up to always use IPv6 when
available. Even my Google Nexus One can use IPv6, when connected to my
WiFi. All my IPv6 capable devices have their own IPv6 address, which
can be reached from the internet. While I have an IPv6 subnet on my
home network, I can also run the 6in4 tunnelling client in single
address mode, on my notebook, when away from home.

BTW, my firewall is OpenSUSE 11.3 running on an old computer.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Carlos E. R.
It is not our decision to make. It is our providers (the ISPs) who must
make that move.

No, you can make the move right now, the same way I did. You use a service
called a "tunnel broker".

What for? What do I gain?
All sites I use are on IPv4.

Look, my printer has IPv6. I can not even connect to it on that! It could
be well that I am clueless.

It says:

TCP/IP(v6)
Status: Ready

Link-Local address: FE80::21E:BFF:FE08:4CCB
Stateless (from Router): Not configured
Stateful (from DHCPv6): Not configured

But if I write in firefox:

http://FE80::21E:BFF:FE08:4CCB

I'm refered to a google page instead of my printer. So, linux is not IPv6
aware >:-)


Yes, I'm playing thick-head here. Just to show you that I have no idea of
IPv6 - and I'm a trained network installer, I have a Cisco diploma
somewhere >:-p

(we saw nothing of IPv6 in the training)

With a tunnel broker, you set up your computer or
firewall to use 6in4 tunnelling. This adds a 20 byte IPv4 header to the IPv6
packet, for transport over the existing IPv4 network. The tunnel broker then
removes that header and puts your IPv6 packet on the IPv6 internet. You can
do this with Linux, Windows, Mac and other. There are also many routers that
support this.

Which I would have to pay. :-(
No, thanks.

Cisco routers certainly do, as do certain D-link models.

Cisco for a home ADSL? Way over the roof in price.


No, let my ISP do the change, they will tell me what I have to do. And
possibly they will pay for the change, being their "fault" >:-)

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0kUpIACgkQtTMYHG2NR9VV7gCfeHoc8tZpTzR/bdkFhUmtOTw9
W3cAn19TAjkfS/1EKkwUc+3ZkzY49Opz
=eULb
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.

Post by James Knott

Post by Carlos E. R.
It is not our decision to make. It is our providers (the ISPs) who
must make that move.

No, you can make the move right now, the same way I did. You use a
service called a "tunnel broker".

What for? What do I gain?
All sites I use are on IPv4.
Look, my printer has IPv6. I can not even connect to it on that! It
could be well that I am clueless.
TCP/IP(v6)
Status: Ready
Link-Local address: FE80::21E:BFF:FE08:4CCB
Stateless (from Router): Not configured
Stateful (from DHCPv6): Not configured
http://FE80::21E:BFF:FE08:4CCB
I'm refered to a google page instead of my printer. So, linux is not
IPv6 aware >:-)

Try this instead:

http://[FE80::21E:BFF:FE08:4CCB]

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Per Jessen

Post by Carlos E. R.
http://FE80::21E:BFF:FE08:4CCB
I'm refered to a google page instead of my printer. So, linux is not
IPv6 aware >:-)

http://[FE80::21E:BFF:FE08:4CCB]

Same thing :-(


***@Telcontar:~> ping6 "FE80::21E:BFF:FE08:4CCB"
connect: Invalid argument
***@Telcontar:~> ping6 "FE80\:\:21E\:BFF\:FE08\:4CCB"
unknown host

***@Telcontar:~> ping6 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.027 ms

***@Telcontar:~> ping6 FE80::21E:BFF:FE08:4CCB
connect: Invalid argument


I have no idea how one is supposed to use this. however, ntop does see
"ff02::1:2", with the MAC address of the printer. There is connectivity on
IPv6. I think it is a dhcpv6-client request from the printer.

I also see activity on ff02::fb.

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0kb9IACgkQtTMYHG2NR9Xd7QCdFg+yCDJJGuUTnm9fPVQQO4o7
DJcAoIz6gTYyM8iTuiViecrNT8ATIvkb
=P9k7
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.
http://FE80::21E:BFF:FE08:4CCB
I'm refered to a google page instead of my printer. So, linux is not
IPv6 aware >:-)

http://[FE80::21E:BFF:FE08:4CCB]

Same thing :-(
connect: Invalid argument
unknown host

Yes, the link-local address can't be used like that - TBH, I don't know
why.
For ping6, you need to specify the interface:

ping6 -I ethX FE80::21E:BFF:FE08:4CCB

[Joachim Schrod]

Post by Per Jessen

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.
http://FE80::21E:BFF:FE08:4CCB
I'm refered to a google page instead of my printer. So, linux is not
IPv6 aware >:-)

http://[FE80::21E:BFF:FE08:4CCB]

Same thing :-(
connect: Invalid argument
unknown host

Yes, the link-local address can't be used like that - TBH, I don't know
why.
ping6 -I ethX FE80::21E:BFF:FE08:4CCB

And that shall be _better_ than plain ol' ping?

To all the IPv6 proponents here: Don't tell users what one *can*
technically do. We've read it a thousand times, it's getting
annoying. Tell users what the advantage of switching to IPv6 is
*now* for home or SOHO scenarios. What does one get by using a
tunnel, besides experience? And while getting experience might be
of interest for network engineers or people who admin networks at
large institutions, it doesn't mean a lot for the majority of
openSUSE users. They didn't need experience for IPv4 -- why should
they need experience with IPv6 at once? If they really would need
it, it would be proof that IPv6 is really much worse than IPv6. So
``getting experience'' is not a valid reason to switch now.

Oh, and please don't tell ``all will be good and everything
works'', that ain't the case. If you do so, you're a fanboy, and
don't act professional. Tell about the problems that one enounters
during switching and how they can be solved.

Yes, we'll have to go to IPv6 eventually, but the user experience
now is just crap.
http://arstechnica.com/business/news/2010/09/there-is-no-plan-b-why-the-ipv4-to-ipv6-transition-will-be-ugly.ars
has a nice and adequate representation. The transition once was
planned to go via dual-stack for several years; but this plan has
gone down the drain, obviously.

Joachim

PS: FTR, I know a lot about IPv6, but then that's part of my job --
my company does network consulting. But I don't think that's
necessary for home and SOHO users, or even for sysadmins in larger
companies that don't have to manage networks. And no, I can't see
any reason at all why people like Carlos should switch before their
ISP offers them full IPv6 connectivity. I surely won't do in my
home network.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Joachim Schrod
To all the IPv6 proponents here: Don't tell users what one *can*
technically do. We've read it a thousand times, it's getting
annoying. Tell users what the advantage of switching to IPv6 is
*now* for home or SOHO scenarios.

Just so.

Post by Joachim Schrod
PS: FTR, I know a lot about IPv6, but then that's part of my job --
my company does network consulting. But I don't think that's
necessary for home and SOHO users, or even for sysadmins in larger
companies that don't have to manage networks. And no, I can't see
any reason at all why people like Carlos should switch before their
ISP offers them full IPv6 connectivity. I surely won't do in my
home network.

Yep.
I'm curious about IPv6, but if I can't even connect to my printer, I don't
see that I gain anything. It may be because I know nothing about it, but
that will be the case for most people.

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lL68ACgkQtTMYHG2NR9Vt6wCeMho84qqKAcQf8b40/UzGhJmF
wVYAn3DrgaIpJrWVZzlDr8YWNrqipUPQ
=48Kc
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Joachim Schrod

Post by Per Jessen
Yes, the link-local address can't be used like that - TBH, I don't
know why.
ping6 -I ethX FE80::21E:BFF:FE08:4CCB

And that shall be _better_ than plain ol' ping?

Nope, I wasn't suggesting anything like that.

Post by Joachim Schrod
To all the IPv6 proponents here: Don't tell users what one *can*
technically do. We've read it a thousand times, it's getting
annoying. Tell users what the advantage of switching to IPv6 is
*now* for home or SOHO scenarios.

There aren't any. I think it must quite clear to most people that the
plain user of an internet connection couldn't care less whether it is
IPv4, IPv6 or RFC1149.

[Per Jessen]

Post by Carlos E. R.
I'm curious about IPv6, but if I can't even connect to my printer, I
don't see that I gain anything. It may be because I know nothing about
it, but that will be the case for most people.

You will only ever gain something if you try to solve the problem. Once
you've solved it, you'll know a lot more.

[Hans Witvliet]

Post by Carlos E. R.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Joachim Schrod
To all the IPv6 proponents here: Don't tell users what one *can*
technically do. We've read it a thousand times, it's getting
annoying. Tell users what the advantage of switching to IPv6 is
*now* for home or SOHO scenarios.

Just so.

Post by Joachim Schrod
PS: FTR, I know a lot about IPv6, but then that's part of my job --
my company does network consulting. But I don't think that's
necessary for home and SOHO users, or even for sysadmins in larger
companies that don't have to manage networks. And no, I can't see
any reason at all why people like Carlos should switch before their
ISP offers them full IPv6 connectivity. I surely won't do in my
home network.

Yep.
I'm curious about IPv6, but if I can't even connect to my printer, I don't
see that I gain anything. It may be because I know nothing about it, but
that will be the case for most people.
- --

To be honest, the usage boils down different, depending how you use the
Net... (content provider vs content user)

Governamental organisations (usa, in) are forced to have their
information they distribute on V6. For usa it is for both v4 and v6.
But in india they are not allowed to start new services on v4.....

So _if_ you have any connections with organisations in india, you might
already be confronted with "unreachable" sites.

otoh, if you are a contractor doing business for organisations in either
of those countries, you will simply be forced, or loose contracts.

If neither applies to you, well, there is is enough sand around ;->

It certainly looks farfatched to have an IPv6-enabled printer, at least
for the end user. But otoh, for a large organisation with printers that
otherwise would be natted, for a service point-of-view it might be handy
to be able to reach them all direcly. But that probably doesn't apply to
you, i presume ;-)

For us, mere mortals in the old world, where ripencc has more than
enough reserves, we just have time to get prepared.
Just like you wrote yourself, and i quote:

""It may be because I know nothing about it, but that will be the case
for most people.""

[James Knott]

Post by Carlos E. R.
Yep.
I'm curious about IPv6, but if I can't even connect to my printer, I
don't see that I gain anything. It may be because I know nothing about
it, but that will be the case for most people.

IPv6 won't give a lot for accessing your printer. IPv4 local addresses
do just fine for that. However, the issue is there are simply not
enough IPv4 addresses to meet todays needs, particularly with 4G phones
on the way. They'll be VoIP only and require a real IP address to work
properly. Of course servers etc. will need addresses and there are
already plenty of people in this world who cannot get an IPv4 address
for home or work. Clinging to IPv4 only makes that problem worse.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Hans Witvliet
To be honest, the usage boils down different, depending how you use the
Net... (content provider vs content user)

...

That is the case. I don't have, as far as I know, to connect to any site
that is on IPv6 only. If I had to, I guess I would learn how to use a
tunnel. When that becomes more common, I hope that ISPs around the world
start providing tunnels, gateways, routers, whatever. They will be forced
to, I hope.

Post by Hans Witvliet
It certainly looks farfatched to have an IPv6-enabled printer, at least
for the end user.

I was simply attempting to try v6 "in house", it is the only thing I have
that I know has IPv6, besides this linux.

Post by Hans Witvliet
But otoh, for a large organisation with printers that
otherwise would be natted, for a service point-of-view it might be handy
to be able to reach them all direcly. But that probably doesn't apply to
you, i presume ;-)

No, it doesn't :-)

Post by Hans Witvliet
For us, mere mortals in the old world, where ripencc has more than
enough reserves, we just have time to get prepared.
""It may be because I know nothing about it, but that will be the case
for most people.""

:-)

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAk0l+HYACgkQtTMYHG2NR9W4XQCgmHg1v3FYN2aSIK7FlxZUY7KY
rG8AnR1YKAGxt7jKuW5CeKyGeiJQG2Ni
=SESo
-----END PGP SIGNATURE-----

[James Knott]

Post by Hans Witvliet
It certainly looks farfatched to have an IPv6-enabled printer

It'd be easy enough to access that printer, if he had an IPv6 subnet set
up. Using the link local address makes things a bit more difficult.

[Hans Witvliet]

Post by Carlos E. R.
What for? What do I gain?
All sites I use are on IPv4.

You probably mean:
"All the sites i used _untill now_, are on IPv4"
Did you ever consider the possibility that there might me a site
tomorrow that is ipv6 only?

Post by Carlos E. R.
Look, my printer has IPv6. I can not even connect to it on that! It could
be well that I am clueless.

That's for the brave at heart.

Post by Carlos E. R.
I'm refered to a google page instead of my printer. So, linux is not IPv6
aware >:-)

Then you probably use SuSE_6.3 or even much older...

Post by Carlos E. R.
Yes, I'm playing thick-head here. Just to show you that I have no idea of
IPv6 - and I'm a trained network installer, I have a Cisco diploma
somewhere >:-p
(we saw nothing of IPv6 in the training)

Well, it is even in the current BSDA-exams

Post by Carlos E. R.

With a tunnel broker, you set up your computer or
firewall to use 6in4 tunnelling. This adds a 20 byte IPv4 header to the IPv6
packet, for transport over the existing IPv4 network. The tunnel broker then
removes that header and puts your IPv6 packet on the IPv6 internet. You can
do this with Linux, Windows, Mac and other. There are also many routers that
support this.

Which I would have to pay. :-(

If you are into Cisco, you wil probably be overpaid...

Post by Carlos E. R.
No, thanks.

Cisco routers certainly do, as do certain D-link models.

Cisco for a home ADSL? Way over the roof in price.
No, let my ISP do the change, they will tell me what I have to do. And
possibly they will pay for the change, being their "fault" >:-)
- --

Well i guess that your v.23 modem probably doesn't support it ;-)

[Per Jessen]

Post by David C. Rankin

Post by Per Jessen

Post by David C. Rankin
Guys,
What is the proper entry for 127.0.0.1 in /etc/hosts?

127.0.0.1 localhost

Post by David C. Rankin
I have seen discussion that 127.0.0.2 is deprecated

Like John says, 127.0.0.0/8 are usually assigned to the loopback
interface, and are usually = localhost.

Well,
IP_address canonical_hostname [aliases...]
showing you can have any number of aliases for a given IP_address
canonical_hostname pair. So I have always included the hostname as an
127.0.0.1 localhost phoenix
which provides resolution for localhost as well as your hostname on
loopback. I have run across several web apps that want to be able to
resolve the loopback address with the hostname. The 127.0.0.2 SuSE has
used is a hack and there is talk about doing away with it completely.

I can't quite figure out what that is good for, and I've certainly never
used it. All of my hostnames resolve to the primary IP-address of the
host.

Post by David C. Rankin
The real problem is with dnsdomainname (or hostname -d) which will not
resolve to anything unless you include the full canonical_hostname in
127.0.0.1 providence.rlfpllc.com localhost providence

It works fine on my systems - dnsdomainname and hostname -d both return
the domain-name.

[David Haller]

Hello,

On Sun, 02 Jan 2011, David C. Rankin wrote:
[..]

Post by David C. Rankin
Well,
IP_address canonical_hostname [aliases...]

[..]

Post by David C. Rankin
The real problem is with dnsdomainname (or hostname -d) which will not resolve
127.0.0.1 providence.rlfpllc.com localhost providence

And that's what you should use. I use the pattern:

127.0.0.1 host.domain.tld localhost host more_aliases
127.0.0.2 www.host.domain.tld www ### for use with apache e.g.
127.0.0.3 docs.host.domain.tld docs ### for use with apache e.g.
[..]
127.0.1.1 ad.doubleclick.com ad.doubleclick.net ad.uk.doubleclick.net
127.0.1.1 ad-emea.doubleclick.net
127.0.1.2 www.google-analytics.com ssl.google-analytics.com google-analytics.com
[..]

If 'hostname' and 'hostname -f' return the correct values, you're good.

I had problems when all adserver/tracker stuff was on one IP (more
than 40 or whatever it was).

-dnh

[John Andersen]

Post by David C. Rankin
127.0.0.1 localhost phoenix
which provides resolution for localhost as well as your hostname on loopback. I
have run across several web apps that want to be able to resolve the loopback
address with the hostname. The 127.0.0.2 SuSE has used is a hack and there is
talk about doing away with it completely.

The 127.0.0.2 is not just specific to opensuse, its been in use since the Pleistocene
in several distros. I suspect it might have been necessary at one time or another
in the early days.

I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).

Any links to such a discussion?

[John Andersen]

Post by John Andersen

Post by David C. Rankin
127.0.0.1 localhost phoenix
which provides resolution for localhost as well as your hostname on loopback. I
have run across several web apps that want to be able to resolve the loopback
address with the hostname. The 127.0.0.2 SuSE has used is a hack and there is
talk about doing away with it completely.

The 127.0.0.2 is not just specific to opensuse, its been in use since the Pleistocene
in several distros. I suspect it might have been necessary at one time or another
in the early days.
I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).
Any links to such a discussion?

https://features.opensuse.org/308824
http://www.njabl.org/use.html
Many more threads about...

Ah, well the first is simply a local suse issue
and the second is actually a valid use of 127/8 for the specific
purpose it was intended for, resolving all sorts of things right back
to the local machine.

There is no inherent reason that njabl would need many such entries,
one would do just the same, and they could have used 127.0.0.1
for all of them. Its simply a book keeping feature, but they
all resolve to local host.

So the second is actually an argument multiple uses of 127.0.0.x, but
nothing they do requires an entry in hosts file, because they rely
on the fact that the entire 128/8 block always points to localhost.

[David C. Rankin]

Post by John Andersen

Post by David C. Rankin
127.0.0.1 localhost phoenix
which provides resolution for localhost as well as your hostname on loopback. I
have run across several web apps that want to be able to resolve the loopback
address with the hostname. The 127.0.0.2 SuSE has used is a hack and there is
talk about doing away with it completely.

The 127.0.0.2 is not just specific to opensuse, its been in use since the Pleistocene
in several distros. I suspect it might have been necessary at one time or another
in the early days.
I have not seen any talk about doing away with it, but the shortage of IP space
would certainly apply pressure in that direction (reducing localhost to a /24 instead
of a /8).
Any links to such a discussion?

Primarily:
https://features.opensuse.org/308824


Others:
http://www.njabl.org/use.html

Many more threads about...
--
David C. Rankin, J.D.,P.E.
Rankin Law Firm, PLLC
510 Ochiltree Street
Nacogdoches, Texas 75961
Telephone: [phone]
Facsimile: [phone]
www.rankinlawfirm.com

[James Knott]

Post by Carlos E. R.
Which I would have to pay.

Not with the tunnel brokers I'm aware of.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Per Jessen
Yes, the link-local address can't be used like that - TBH, I don't know
why.
ping6 -I ethX FE80::21E:BFF:FE08:4CCB

Ah, yes, I remember that now. As I don't use v6 and I can't understand the
reason of having to specify the interface, so I always forget that detail.

Yes, that ping works, and then the address appears in ntop. The syntax for
browsing is, or so says ntop, "http://[ff02::1:ff08:4ccb]/", I don't know
why is the difference - at least, that's what ntop says it is. But it
doesn't work:

Firefox can't establish a connection to the server at [ff02::1:ff08:4ccb].

* The site could be temporarily unavailable or too busy. Try again in a few
moments.

* If you are unable to load any pages, check your computer's network
connection.

* If your computer or network is protected by a firewall or proxy, make sure
that Firefox is permitted to access the Web.

But the printer does not respond to that IP by ping, either; it does to
"FE80::21E:BFF:FE08:4CCB" instead. However, if I put
"http://[FE80::21E:BFF:FE08:4CCB]/" in FF, FF doesn't even try to connect,
I keep seeing the same page that was there previously.


Obviously, I don't know if the printer is properly configured, or if my
linux is. I guess it is, as I can ping the printer. But I have no idea.


I also see entries in the firewall:


<0.4> 2011-01-06 03:18:13 Telcontar kernel - - - [256669.247433]
SFW2-INext-DROP-DEFLT IN=eth0 OUT=
MAC=00:21:85:16:2d:0b:00:1e:0b:08:4c:cb:86:dd
SRC=fe80:0000:0000:0000:021e:0bff:fe08:4ccb
DST=fe80:0000:0000:0000:0221:85ff:fe16:2d0b LEN=104 TC=0 HOPLIMIT=1
FLOWLBL=0 PROTO=ICMPv6 TYPE=129 CODE=0 ID=62255 SEQ=1

<0.4> 2011-01-06 03:18:14 Telcontar kernel - - - [256670.247202]
SFW2-INext-DROP-DEFLT IN=eth0 OUT=
MAC=00:21:85:16:2d:0b:00:1e:0b:08:4c:cb:86:dd
SRC=fe80:0000:0000:0000:021e:0bff:fe08:4ccb
DST=fe80:0000:0000:0000:0221:85ff:fe16:2d0b LEN=104 TC=0 HOPLIMIT=1
FLOWLBL=0 PROTO=ICMPv6 TYPE=129 CODE=0 ID=62255 SEQ=2

<0.4> 2011-01-06 03:18:15 Telcontar kernel - - - [256671.247229]
SFW2-INext-DROP-DEFLT IN=eth0 OUT=
MAC=00:21:85:16:2d:0b:00:1e:0b:08:4c:cb:86:dd
SRC=fe80:0000:0000:0000:021e:0bff:fe08:4ccb
DST=fe80:0000:0000:0000:0221:85ff:fe16:2d0b LEN=104 TC=0 HOPLIMIT=1
FLOWLBL=0 PROTO=ICMPv6 TYPE=129 CODE=0 ID=62255 SEQ=3

<0.4> 2011-01-06 03:18:16 Telcontar kernel - - - [256672.247220]
SFW2-INext-DROP-DEFLT IN=eth0 OUT=
MAC=00:21:85:16:2d:0b:00:1e:0b:08:4c:cb:86:dd
SRC=fe80:0000:0000:0000:021e:0bff:fe08:4ccb
DST=fe80:0000:0000:0000:0221:85ff:fe16:2d0b LEN=104 TC=0 HOPLIMIT=1
FLOWLBL=0 PROTO=ICMPv6 TYPE=129 CODE=0 ID=62255 SEQ=4

<0.4> 2011-01-06 03:18:17 Telcontar kernel - - - [256673.247210]
SFW2-INext-DROP-DEFLT IN=eth0 OUT=
MAC=00:21:85:16:2d:0b:00:1e:0b:08:4c:cb:86:dd
SRC=fe80:0000:0000:0000:021e:0bff:fe08:4ccb
DST=fe80:0000:0000:0000:0221:85ff:fe16:2d0b LEN=104 TC=0 HOPLIMIT=1
FLOWLBL=0 PROTO=ICMPv6 TYPE=129 CODE=0 ID=62255 SEQ=5


- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lKQEACgkQtTMYHG2NR9V41gCfatlaI3hL7ffAHRtEoc4odtBm
XmQAnjlKEjyo6vnvxcs+bmEXdQG/BRlO
=Oqss
-----END PGP SIGNATURE-----

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Hans Witvliet

Post by Carlos E. R.
What for? What do I gain?
All sites I use are on IPv4.

"All the sites i used _untill now_, are on IPv4"
Did you ever consider the possibility that there might me a site
tomorrow that is ipv6 only?

That will be tomorrow. Or beyond. And tomorrow, it will still be tomorrow. >:-)

Post by Hans Witvliet

Post by Carlos E. R.
I'm refered to a google page instead of my printer. So, linux is not IPv6
aware >:-)

Then you probably use SuSE_6.3 or even much older...

Last time I looked it was 11.2 at least.

Post by Hans Witvliet

Post by Carlos E. R.
(we saw nothing of IPv6 in the training)

Well, it is even in the current BSDA-exams

Not in what we did.

It was a training course over 250 hours paid by the government. If the
government thinks I should know nothing about IPv6, who am I to argue? >:-)

Post by Hans Witvliet

Post by Carlos E. R.
Which I would have to pay. :-(

If you are into Cisco, you wil probably be overpaid...

Ha! Wish I were. Any of that, Cisco or overpaid. I would be grateful to
be paid at all.

Post by Hans Witvliet
Well i guess that your v.23 modem probably doesn't support it ;-)

I guess it does support IPv6. It is only transmission hardware, it knows
nothing of the IP. That's handled by the ppp daemon in Linux... so, is the
ppp Linux daemon IPv6 capable? Einn? >:-)

- --
Cheers,
Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0lRY0ACgkQtTMYHG2NR9VxPQCbBevyV2AnNDaHIwYWDbE0rOYU
YggAn2Wl5wYGCKDdPMhsBS1tQwjJdj+E
=TWc3
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.
Firefox can't establish a connection to the server at
[ff02::1:ff08:4ccb].

Try this instead:

http:///[FE80::21E:BFF:FE08:4CCB%eth0]

That format works for me.

[Per Jessen]

Post by Per Jessen

Post by Carlos E. R.
Firefox can't establish a connection to the server at
[ff02::1:ff08:4ccb].

http:///[FE80::21E:BFF:FE08:4CCB%eth0]
That format works for me.

Correction: http://[FE80::21E:BFF:FE08:4CCB%eth0]

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Per Jessen

Post by Carlos E. R.
Firefox can't establish a connection to the server at
[ff02::1:ff08:4ccb].

http:///[FE80::21E:BFF:FE08:4CCB%eth0]
That format works for me.

Indeed! That one works. Three or two bars, both work.

But it is a terrible format!

(thunderbird doesn't highlight it. It is not clickable)

Having to add a 6 to the ping, specify the interface in pings or web
browsing... Perhaps it is a question of routing tables :-?



What would happen if I modify my bind dns configuration in an unfathomable
way, so that the returned address for the printer is the v6 one instead of
the v4 one, would I have to write the port too? That's absurd. It is, isn't it?

So if I bookmark the address, it would go with eth0 too...

How weird.


And where from did the printer get that address? I did not configure it, it
was HP, the manufacturer. Perhaps I should configure instead the equivalent
v6 address to the v4 addr.

Ha! I'm lucky to have a v4 ISP, or you could be printing to /my/ printer
from your home! >:-)


Is there an IPv6 manual for dummies?

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/

iEYEARECAAYFAk0l+IwACgkQtTMYHG2NR9W+/wCeLauJNMNx7Xgdkix4WbxF37Nz
wOoAnj5e99xjCiZHi2wYfUm8dtnlI8nj
=tj1W
-----END PGP SIGNATURE-----

[Per Jessen]

Post by Carlos E. R.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Per Jessen

Post by Carlos E. R.
Firefox can't establish a connection to the server at
[ff02::1:ff08:4ccb].

http:///[FE80::21E:BFF:FE08:4CCB%eth0]
That format works for me.

Indeed! That one works. Three or two bars, both work.
But it is a terrible format!

Yes, it is unusual having an IP-address in the URL, normally it would be
a name, but the %ethX extension doesn't work with a name (as far as I
have seen). Btw, the FE80::21E:BFF:FE08:4CCB%eth0 format also works
with e.g. ssh and telnet. It is handled by the getaddrinfo() function.
I suspect there must be a way of specifying a name plus devicename, but
I haven't found it yet.

Post by Carlos E. R.
Having to add a 6 to the ping, specify the interface in pings or web
browsing... Perhaps it is a question of routing tables :-?

It is a problem specific to link-local addresses - if you were to use
global unicast addresses, i.e. "plain" IPv6 addresses, there is no
problem. If you can assign your own IPv6 addresses to the printer, you
could try using the fec0 prefix - it's deprecated, but was meant to be
site-local.

Post by Carlos E. R.
What would happen if I modify my bind dns configuration in an
unfathomable way, so that the returned address for the printer is the
v6 one instead of the v4 one, would I have to write the port too?
That's absurd. It is, isn't it?

If you change the A record and make it an AAAA instead with a proper
IPv6 address, it'll work just fine. I tried it with a link-local
address, and I couldn't get it to work with the %ethX notation.

Post by Carlos E. R.
And where from did the printer get that address? I did not configure
it, it was HP, the manufacturer. Perhaps I should configure instead
the equivalent v6 address to the v4 addr.

It's automatically configured; the fe80 prefix means "link-local", the
rest is an encoding of MAC-address of the interface.

Post by Carlos E. R.
Is there an IPv6 manual for dummies?

I'm sure there are many good books out there, but I can't recommend
anyone specific. Sofar I've managed with a couple good wikipedia
articles and googling a lot.

[Carlos E. R.]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Resending, lets see if Telefonica behaves properly now.

Post by Per Jessen

Post by Carlos E. R.
But it is a terrible format!

Yes, it is unusual having an IP-address in the URL, normally it would be
a name, but the %ethX extension doesn't work with a name (as far as I
have seen). Btw, the FE80::21E:BFF:FE08:4CCB%eth0 format also works
with e.g. ssh and telnet. It is handled by the getaddrinfo() function.
I suspect there must be a way of specifying a name plus devicename, but
I haven't found it yet.

I meant terrible compared to an IPv4 numerical address. There is no need
for brackets nor for the interface.

Post by Per Jessen

Post by Carlos E. R.
Having to add a 6 to the ping, specify the interface in pings or web
browsing... Perhaps it is a question of routing tables :-?

It is a problem specific to link-local addresses - if you were to use
global unicast addresses, i.e. "plain" IPv6 addresses, there is no
problem. If you can assign your own IPv6 addresses to the printer, you
could try using the fec0 prefix - it's deprecated, but was meant to be
site-local.

Well, I certainly want local-only addresses, not global addresses. I read
somewhere that old v4 addresses could be mapped to new v6 addresses. Some
prefix, then the old 192.168...

(I need a good read on all this...)

Post by Per Jessen

Post by Carlos E. R.
What would happen if I modify my bind dns configuration in an
unfathomable way, so that the returned address for the printer is the
v6 one instead of the v4 one, would I have to write the port too?
That's absurd. It is, isn't it?

If you change the A record and make it an AAAA instead with a proper
IPv6 address, it'll work just fine. I tried it with a link-local
address, and I couldn't get it to work with the %ethX notation.

No, I don't want to put the eth part :-)

So add an AAAA record. But also add the reverse file somehow, too.

Post by Per Jessen

Post by Carlos E. R.
And where from did the printer get that address? I did not configure
it, it was HP, the manufacturer. Perhaps I should configure instead
the equivalent v6 address to the v4 addr.

It's automatically configured; the fe80 prefix means "link-local", the
rest is an encoding of MAC-address of the interface.

Ahh! I see.


And I don't need routing tables, because we write the eth0 port in the
address given to firefox. If I would be using names, how would the routing
know on which ether port is that MAC address? Routing tables are done by
prefixes common to many machines. Unless the operating system does
discovery and automatic routing...

Post by Per Jessen

Post by Carlos E. R.
Is there an IPv6 manual for dummies?

I'm sure there are many good books out there, but I can't recommend
anyone specific. Sofar I've managed with a couple good wikipedia
articles and googling a lot.

I'm more lazy than that, I would prefer a book in paper, or a really good,
simple, web page. :-)

- --
Cheers / Saludos,

Carlos E. R.
(from 11.2 x86_64 "Emerald" at Telcontar)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.12 (GNU/Linux)

iEYEARECAAYFAk0mJDIACgkQtTMYHG2NR9XmwQCdFtkZUWtg41pZvYW0ser1BYMq
CIAAnReOHHL5JnueDtCESgVGWgyOXapi
=wMgo
-----END PGP SIGNATURE-----

[James Knott]

Post by Per Jessen
Yes, the link-local address can't be used like that - TBH, I don't know
why.
ping6 -I ethX FE80::21E:BFF:FE08:4CCB

The reason you have to specify the interface for the link local address,
is because otherwise the computer won't know what interface to use (I
have 3 NICs in one computer). With other addresses, the interface can
be determined by the subnet the address belongs to. Since the link
local address isn't in a subnet, that method doesn't work. However, the
link local address is used to access routers on your network. In that
case, the router periodically announces itself to the computers on the
local network and from that, the computers know which interface the
router is connected to. Here's an example from my computer showing the
default IPv6 route:

default via fe80::202:a5ff:fe7b:d908 dev eth0 proto kernel metric
1024 expires 1695sec mtu 1280 advmss 1220 hoplimit 64

Notice that it specifies eth0 as the interface to find the router.

[James Knott]

Post by Per Jessen

Post by Carlos E. R.
Firefox can't establish a connection to the server at
[ff02::1:ff08:4ccb].

http:///[FE80::21E:BFF:FE08:4CCB%eth0]
That format works for me.

That's what I was trying to remember. Since I have a proper IPv6
subnet, I don't hae to worry about using the link local addresses. I
can just use the address by itself, without specifying the interface. I
can even use the host name, as my DNS server hands out IPv6 & IPv4
addresses.

BTW, you can also put the host names in the hosts file. If you put the
IPv6 line in ahead of the IPv4 line, your apps will use IPv6 when
available. In the hosts file on the computer running dnsmask, the hosts
file contains one block of lines with the IPv6 addresses for the various
host names and then a block for IPv4 addresses. Both sections use the
same host name for each device. Of course the IPv4 section has a couple
more lines for IPv4 only devices.

[James Knott]

Post by Carlos E. R.
And where from did the printer get that address? I did not configure it, it
was HP, the manufacturer. Perhaps I should configure instead the equivalent
v6 address to the v4 addr.

Every IPv6 capable device automatically gets a link local address that
starts out with FE80 and includes the mac address. However, the mac
address is split into two parts, the first 3 bytes, which correspond to
the manufacturer, then FFFE, then the last 3 bytes, which are the
manufacturers sequential number for the NIC. The FFFE is used to fill
out the 48 bit mac address to the 64 bit subnet address. (The smallest
permissible IPv6 subnet is /64.)

[James Knott]

Post by Carlos E. R.
What would happen if I modify my bind dns configuration in an unfathomable
way, so that the returned address for the printer is the v6 one instead of
the v4 one, would I have to write the port too? That's absurd. It is, isn't it?

A valid IPv6 address is something like this: ::ffff:172.16.1.10. You
just add the ::ffff: to the IPv4 address. However, your operating
system recognizes it for what it is and contacts it as IPv4. You only
have to specify the port when using the link local address. If you had
a subnet, even one not connected to the internet, your computer could
figure out which interface to use. Just like in IPv4, there are address
blocks for that purpose.

[James Knott]

Post by Carlos E. R.
Is there an IPv6 manual for dummies?

Actually, O'Reilly has a good book. It's "IPv6 Essentials" IIRC.

[Per Jessen]

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.
But it is a terrible format!

Yes, it is unusual having an IP-address in the URL, normally it would
be a name, but the %ethX extension doesn't work with a name (as far
as I
have seen). Btw, the FE80::21E:BFF:FE08:4CCB%eth0 format also works
with e.g. ssh and telnet. It is handled by the getaddrinfo()
function. I suspect there must be a way of specifying a name plus
devicename, but I haven't found it yet.

I meant terrible compared to an IPv4 numerical address. There is no
need for brackets nor for the interface.

In the normal case, that would be the same for IPv6. The numerical
address is too long to be really useful by itself. Using the IPv6
numerical address is an exception, not the norm.

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.
Having to add a 6 to the ping, specify the interface in pings or web
browsing... Perhaps it is a question of routing tables :-?

It is a problem specific to link-local addresses - if you were to use
global unicast addresses, i.e. "plain" IPv6 addresses, there is no
problem. If you can assign your own IPv6 addresses to the printer,
you could try using the fec0 prefix - it's deprecated, but was meant
to be site-local.

Well, I certainly want local-only addresses, not global addresses.

Well, as far as I've understood, that's actually not foreseen in IPv6.
There is a concept of site-local addresses, but it's been deprecated.
IPv6 has enough addresses for everyone.

Post by Carlos E. R.
I read somewhere that old v4 addresses could be mapped to new v6
addresses. Some prefix, then the old 192.168...

Yes, that is true - one of the transition mechanisms is 6to4, where all
IPv4 addresses are mapped into IPv6 as 2002:: - doesn't mean the
private RFC1918 addresses suddenly become public though.

Post by Carlos E. R.

Post by Per Jessen

Post by Carlos E. R.
What would happen if I modify my bind dns configuration in an
unfathomable way, so that the returned address for the printer is
the v6 one instead of the v4 one, would I have to write the port
too? That's absurd. It is, isn't it?

If you change the A record and make it an AAAA instead with a proper
IPv6 address, it'll work just fine. I tried it with a link-local
address, and I couldn't get it to work with the %ethX notation.

No, I don't want to put the eth part :-)
So add an AAAA record. But also add the reverse file somehow, too.

You probably don't have to, but it's just as easy as an IPv4 reverse
map.

Post by Carlos E. R.
And I don't need routing tables, because we write the eth0 port in the
address given to firefox. If I would be using names, how would the
routing know on which ether port is that MAC address?

The same it does today for IPv4 - it has a route.

Post by Carlos E. R.
Routing tables are done by prefixes common to many machines. Unless
the operating system does discovery and automatic routing...

It does, just like today.

[Per Jessen]

Post by James Knott

Post by Per Jessen
Yes, the link-local address can't be used like that - TBH, I don't
know why.
ping6 -I ethX FE80::21E:BFF:FE08:4CCB

The reason you have to specify the interface for the link local
address, is because otherwise the computer won't know what interface
to use (I have 3 NICs in one computer).

How about a simple workstation with just one NIC?
Besides, in a box with multiple interfaces, each interface has it's own
link-local address (AFAICT), but for some reason the routing table
shows all link-local addresses accessible via lo(opback). There must
be a reason, I just don't know what it is.

[James Knott]

Post by Per Jessen

Post by Carlos E. R.
I meant terrible compared to an IPv4 numerical address. There is no

Post by Carlos E. R.
need for brackets nor for the interface.

In the normal case, that would be the same for IPv6. The numerical
address is too long to be really useful by itself. Using the IPv6
numerical address is an exception, not the norm.

The reason you need the brackets is because of the colons used to
separate the hex digits. The colon is also used to specify a port
number, so you need some way to separate the address from port.