[radvd-devel-l] New option idea: Base6Interface

[Reuben Hawkins]

On Tue, Mar 1, 2011 at 1:35 PM, Stephen Dawkins <elfarto at elfarto.com> wrote:

> On 26/02/2011 17:13, Reuben Hawkins wrote:
>
>> Hi Stephen,
>>
>> Base6Interface is committed to CVS.  It's ready for a test drive.
>>
>> Thanks,
>> Reuben
>>
>>
> Hi Reuben
>
> Thanks for the quick response. I finally got around to testing it. It works
> fine if you specify the prefix as 2000::/64, but not if you specify the
> prefix as ::/64.
>
> Log snippets:
>
> with ::/64
> [Mar 01 21:30:51] radvd: version 1.7 started
> [Mar 01 21:30:51] radvd: all-zeros prefix in radvd.conf, line 5, parsing..
> [Mar 01 21:30:51] radvd: no auto-selected prefix on interface eth0,
> disabling advertisements
>
> with 2000::/64
> [Mar 01 21:31:28] radvd: version 1.7 started
> [Mar 01 21:31:28] radvd: using prefixes on interface br0 for prefixes on
> interface eth0
> [Mar 01 21:31:28] radvd: auto-selected prefix 2001:470:1f09:14ab::/64 on
> interface eth0 from interface br0
>
> It seems the all-zero prefix causes other code to run and interferes with
> the Base6Interface code.
>
> Thanks & Regards
> Stephen
>
>

Hi Stephen,

Yes, that is expected.  All zero prefix activates auto-select on the
interface itself.  So...

interface eth0 {
  prefix ::/0;
};

...will select the prefixes from the global unicast addresses already set on
eth0.  Base6Interface masks in whatever prefix you specify with a prefix
from your Base6Interface....   So...

interface eth0 {
  prefix ffff:ffff:ffff:ffff::/64 {
    Base6Interface br0;
  };
};

Would give you a prefix something like...  2001:1234:5678:ffff::/64 assuming
the prefix len on br0 is /48.

IMHO, It's somewhat a kludge.  I think the auto-selecting code should get
prefixes from the routing table, not the actual addresses on the interfaces,
and auto-selecting prefixes shouldn't look so similar to a real prefix (i.e.
prefix ::/64)...but that's a project for another day...

Thanks,
Reuben
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litech.org/pipermail/radvd-devel-l/attachments/20110301/a6c9354c/attachment.html>