ot, configuring ipv6 routing to tap device

ot, configuring ipv6 routing to tap device

[Gregory Nowak]

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

Hello all.

Apologies for the off topic post, but since those happen on and off
here anyway, I thought I'd post my quandary, and see if someone might
be able to help.

Like the subject says, I'm trying to get ipv6 traffic routed to a tap
device. Let me explain in detail. I have a user-mode linux vm running
on a host system. The vm is able to communicate with the host, my lan,
and the rest of the world just fine via ipv4 over a persistent tap
device, tap0. I'm trying to get the vm to use ipv6. I have a routed
prefix, the machines on my lan can talk to each other, and the world
over ipv6, and the world can in turn talk to all of them over ipv6.

For the purposes of my post, let's say I'm using the 2001:db8::/32
address space, though I am in fact using a routed address
space. Here's what I have, tap0 on the host is configured as
2001:db8::1/32, and eth0 inside the vm is configured as
2001:db8::2/32. If I use telnet from inside the vm to try to connect
to the host, my lan, or the world, I get network is unreachable. If I
run tcpdump on the tap0 interface on the host, I can see neighbor
solicitations going out from the vm, but tap0 is as far as those
packets go. If I try to connect to the vm from the host, or my lan, I
get no route to host, and no packets appear on the host's tap0
interface. 

So, it looks like the vm and tap0 are setup just fine, and I need to
figure out how to route ipv6 traffic between the tap
device/host/lan/world. If on the host I do
ip -6 route add 2001:db8::2
I can then connect to all ports on the host system from inside the vm,
but I still can't connect to my lan/world from the vm, and the
host/lan/world still can't connect to the vm, coming back with no
route to host.

During my googling, I've found people with the same issue on the
user-mode-linux-users list, but no solution. This looks like it
definitely should work if configured right, so I'm assuming the people
that had this problem either solved it, and didn't share with that
list, or they just gave up on it. Other pages on the web indicate that
simply assigning an address to the vm, and to the tap device should be
enough to make things work, but it obviously isn't, at least not for
me, and for others as far as I can gather from my research.

So, can someone please explain where I'm going wrong, and how to do it
correctly? If I come across something that works, I'll post back, so
that this is at least documented somewhat out on the web. If more info
is needed, please let me know.

Once I get this working, since there's no need for the tap device
to have a routed ip as far as I can tell, what I'd like to do if
possible, is to in fact really place the tap device, and the vm on the
2001:db8::/32 address space, and also assign a routable ip to eth0 inside
the vm, assuming that is possible, and could be made to work. Thanks
in advance for any help.

Greg


- -- 
web site: http://www.romuald.net.eu.org
gpg public key: http://www.romuald.net.eu.org/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)

- --
Free domains: http://www.eu.org/ or mail dns-manager@EU.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkoPH4IACgkQ7s9z/XlyUyDBhwCg0TlCHfac/Fi8q9DpcSkzwL4V
5pwAoLCM+PXOAqA7qb4gJFgXJkOCjKEf
=z0qn
-----END PGP SIGNATURE-----

Re: ot, configuring ipv6 routing to tap device

[Jason White]

Gregory Nowak  <speakup@braille.uwo.ca> wrote:

>For the purposes of my post, let's say I'm using the 2001:db8::/32
>address space, though I am in fact using a routed address
>space. Here's what I have, tap0 on the host is configured as
>2001:db8::1/32, and eth0 inside the vm is configured as
>2001:db8::2/32. If I use telnet from inside the vm to try to connect
>to the host, my lan, or the world, I get network is unreachable. If I
>run tcpdump on the tap0 interface on the host, I can see neighbor
>solicitations going out from the vm, but tap0 is as far as those
>packets go. 

Do you have tap0 configured in your /etc/radvd.conf file? If you want to send
it out a router advertisement, you'll need to have radvd running and
radvd.conf configured.

The ip command is the one to use if you need to set up static routes - see the
IPv6 HOWTO. You might also need to enable IPv6 forwarding in your kernel on
the host system.

Re: ot, configuring ipv6 routing to tap device

[Gregory Nowak]

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

On Mon, May 18, 2009 at 04:58:15AM +0000, Jason White wrote:
> Do you have tap0 configured in your /etc/radvd.conf file? If you want to send
> it out a router advertisement, you'll need to have radvd running and
> radvd.conf configured.

No, I actually don't want to have the vm configured via router
advertisements. However, I did want to see if having the eth0 card in
the vm be auto configured would make a difference, so I did add tap0
to radvd.conf, and restarted radvd. The vm did get a ipv6 address from
radvd, but that still didn't solve the routing problems, which was
actually what I expected, but I did hope I might be pleasantly surprised.

> 
> The ip command is the one to use if you need to set up static routes - see the
> IPv6 HOWTO.

The ipv6 howto doesn't mention an if command anywhere as far as I
can tell, it does mention the ip command quite frequently though (smile).

> You might also need to enable IPv6 forwarding in your kernel on
> the host system.

Yes, that's been done, and I forgot to mention it in my initial post.

I think I actually figured out a solution though by bridging the eth1
interface for my lan to the vm's tap0 interface on the host. I'll make
the config changes tomorrow, and will report back if it works, though
I think chances are very high that it's going to. We'll see I guess.


- -- 
web site: http://www.romuald.net.eu.org
gpg public key: http://www.romuald.net.eu.org/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)

- --
Free domains: http://www.eu.org/ or mail dns-manager@EU.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkoRBQMACgkQ7s9z/XlyUyAWIACePjjBg9gMZBLCf9Hpw/xTy3jK
StQAn3cg2x7A9Qb/fIpJkML2aVZywcRG
=RdyV
-----END PGP SIGNATURE-----

Re: ot, configuring ipv6 routing to tap device

[Gregory Nowak]

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

On Sun, May 17, 2009 at 11:49:39PM -0700, Gregory Nowak wrote:
> I think I actually figured out a solution though by bridging the eth1
> interface for my lan to the vm's tap0 interface on the host. I'll make
> the config changes tomorrow, and will report back if it works, though
> I think chances are very high that it's going to. We'll see I guess.

Well, turns out I was right, and I now have the uml talking to my lan
and world over ipv4, and ipv6, and my lan and the world talking to the
uml over ipv6. So basically what I did is to install the bridge-utils
package under debian, read the docs, and configure things in
/etc/network/interfaces, all on the host of course. The config on the
uml wasn't touched at all. For a generic config method, the brctl(8)
manual page describes how to set things up. Hth someone else, and
thanks to Jason for his suggestions.

Greg


- -- 
web site: http://www.romuald.net.eu.org
gpg public key: http://www.romuald.net.eu.org/pubkey.asc
skype: gregn1
(authorization required, add me to your contacts list first)

- --
Free domains: http://www.eu.org/ or mail dns-manager@EU.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkoSDqkACgkQ7s9z/XlyUyDVtwCdGptbS8XeUkks9O8JSoyQtpEz
YL0AniC/fwQXIs52wuig9ho+AZbtWNMW
=8k6P
-----END PGP SIGNATURE-----