Web lists-archives.com

Re: IPv6 addresses




On 8/02/19 8:08 AM, Pascal Hambourg wrote:
> Le 07/02/2019 à 10:15, Richard Hector a écrit :
>>>
>>> # this one should hopefully configure via SLAAC
>>> iface ens3 inet6 auto
>>>
>>> # this one is for my static stuff (using a ULA prefix)
>>> iface ens3 inet6 static
>>>    address 2001:db8::1/64
>>>    up ip -6 route add ...
>>>    up ip -6 route add ...
>>>    up ip -6 route add ...
>>>
>>> But I don't get the SLAAC address. The only way I've managed to get the
>>> SLAAC address plus the static, is to have a single auto stanza like
>>> this:
>>>
>>> iface ens3 inet6 auto
>>>    up ip -6 addr add 2001:db8::1/64 dev ens3
>>>    up ip -6 route add ...
>>>    up ip -6 route add ...
>>>    up ip -6 route add ...
>>>
>>> Incidentally, that also gives me an autoconfigured address (based on the
>>> MAC) in the 2001:db8::/64 prefix, which I don't think I need.
> 
> Huh ? Isn't it the autoconfigured address you expect to get from an
> "auto" stanza ?

Sorry for the confusion. I expected to get a autoconfigured address in
the prefix the router got from the isp, not in a ULA one I created
myself. However, it's starting to look like the router actually does
advertise the ULA as well (I thought it didn't), so that's probably
correct behaviour.

>>> Does that suggest a bug in ifupdown? Or am I interpreting something
>>> wrong?
> 
> I would call it more an unforeseen consequence.
> 
>> Actually, if I put the 'auto' stanza after the 'static' one, it does
>> work (I still get the autogenerated address in the ULA prefix).
>>
>> Is it supposed to be order-sensitive?
> 
> Yes. "auto" sets autoconf=1 to enable SLAAC, and "static" sets
> autoconf=0 by default, which disables SLAAC and removes any
> autoconfigured address.

I didn't find anywhere in the man page that said the order of my stanzas
was important - I assumed they were entirely independent of each other.

>> I can also make it work by adding 'autoconf 1' to the static stanza, and
>> omitting the auto stanza.
> 
> This is the right way to go.

And it's the way I'm going.

It just appears that the idea of using one stanza per ip address (or at
least per prefix) doesn't seem to work consistently.

Cheers,
Richard


Attachment: signature.asc
Description: OpenPGP digital signature