On 30 Nov 2018, at 15:59, Christoph Moench-Tegeder wrote: > ## O. Hartmann (ohartmann_at_walstatt.org): > >> As far as I know, with the IPv4 stack a IPv4 address is obtained >> automatically, so I would expect the same for IPv6. > > The fun with "automatically" is that there's more than one way... > DHCPv6 and NDP (IPV6 Neighbour Discovery Protocol/Router Solicitation) > have been mentioned, the third option is IPV6CP (PPP options, just as > PPP-with-IPv4 does with IPCP). I've no idea what your provider does, so... No, IPV6CP, to my very best 15 year old memory only negotiates the interface identifiers, which are used to generate the link-local addresses. There is no negotiation for full prefix/global addresses, hence it is different to the IPCP NCP used for IPv4. One wants to run rtsol on the link and then depending on the O/M bits possibly also DHCPv6 I’d assume. There’s a couple of other options and shortcuts, on how to configure global addresses (as always, if you know you have a static prefix assigned, one can do it by hand for example); and then there are shortcuts as to when you’d perform DAD, which shouldn’t bother the user. These days there should also be options with regards to RFC4941 (privacy extensions for stateless address autoconf). It may no be uncommon to run the ptp-link with link-local addresses only and configure an address from a prefix on lo0 or the internal interface only; but I am doubtful FreeBSD’s userland implementation is that sophisticated. /bzReceived on Fri Nov 30 2018 - 15:40:39 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:19 UTC