And here I've made a fool of myself yet again. From the documentation:
> The dhcp Method
> This method may be used to obtain network interface configuration via
> stateful DHCPv6 with dhclient. In stateful DHCPv6, the DHCP server is
> responsible for assigning addresses to clients.
> Options
> accept_ra int
> Accept router advertisements (0=off, 1=on, 2=on+forwarding). Default
> value: "1"
> autoconf int
> Perform stateless autoconfiguration (0=off, 1=on)
Strange how accept_ra was 2 as shown by sysctl, but it still didn't
work. Maybe that autoconf option was the ticket? Didn't see many other
folks trying these options out. Oh well.
Thank you for helping me think this through properly. This has been
solved.
- Neil
On Fri, Nov 15, 2019 at 01:14:39PM -0800, Neil E. Hodges wrote:
> I ended up cheating by having this setup enable both wide-dhcpv6 for the
> IP address, and SLAAC for the routing information. It's really odd how
> the configuration built into the system doesn't allow it.
>
> /etc/network/interfaces
> > auto eth-external
> > iface eth-external inet dhcp
> > pre-up modprobe ipv6
> > post-up /etc/init.d/wide-dhcpv6-client start
> > pre-down /etc/init.d/wide-dhcpv6-client stop
> >
> > iface eth-external inet6 auto
>
> /etc/wide-dhcpv6/dhcp6c.conf
> > interface eth-external {
> > send ia-na 1;
> > send rapid-commit;
> > script "/etc/wide-dhcpv6/dhcp6c-script";
> > };
> >
> > id-assoc na 1 {
> };
>
> Why in the world does enabling DHCPv6 for the IP address cause the
> kernel to ignore router advertisements? That's a common use case for
> IPv6.
>
> Thank you,
>
> - Neil
>
>
> On Thu, Nov 14, 2019 at 07:01:47PM -0800, Neil E. Hodges wrote:
> > Hello everybody,
> >
> > I've been having a bear of a time trying to get my Debian Buster-based
> > router
> > to operate with Comcast's native IPv6 configuration. I'm not looking
> > for prefix delegation, just a public address and a default gateway. I'm
> > using the ISC DHCP client for both v4 and v6 since it's generally worked
> > for me other than this.
> >
> > I've tried two approaches, with eth-external being the name I configured
> > for the interface in udev.
> >
> > 1. iface eth-external inet6 dhcp
> >
> > This gets me an IPv6 address just fine, but no route whatsoever unless I
> > manually configure it. I see the router advertisements coming in when
> > running tcpdump, but the kernel just disregards them.
> >
> > 2. iface eth-external inet6 auto
> > dhcp 1
> > request_prefix 1
> >
> > This usually gets me a /64 prefix and a route, but not an IPv6 address.
> >
> > The usual sysctl suspects don't reveal anything significant:
> >
> > > # sysctl -a | grep -i 'eth-external.*ra'
> > > net.ipv4.conf.eth-external.drop_gratuitous_arp = 0
> > > net.ipv4.neigh.eth-external.retrans_time_ms = 1000
> > > net.ipv6.conf.eth-external.accept_ra = 2
> > > net.ipv6.conf.eth-external.accept_ra_defrtr = 1
> > > net.ipv6.conf.eth-external.accept_ra_from_local = 0
> > > net.ipv6.conf.eth-external.accept_ra_min_hop_limit = 1
> > > net.ipv6.conf.eth-external.accept_ra_mtu = 1
> > > net.ipv6.conf.eth-external.accept_ra_pinfo = 1
> > > net.ipv6.conf.eth-external.accept_ra_rt_info_max_plen = 0
> > > net.ipv6.conf.eth-external.accept_ra_rt_info_min_plen = 0
> > > net.ipv6.conf.eth-external.accept_ra_rtr_pref = 1
> > > net.ipv6.conf.eth-external.dad_transmits = 1
> > > net.ipv6.conf.eth-external.suppress_frag_ndisc = 1
> > > net.ipv6.neigh.eth-external.retrans_time_ms = 1000
> >
> > And no packets are being dropped, as I have rules to allow ICMPv6 and
> > DHCP:
> > > -A INPUT -p ipv6-icmp -j ACCEPT
> > > -A INPUT -s fe80::/10 -i eth-external -p udp -m udp --sport 546 --dport
> > > 547 -j ACCEPT
> > > -A INPUT -s fe80::/10 -i eth-external -p udp -m udp --sport 547 --dport
> > > 546 -j ACCEPT
> >
> > Nothing is logged matching those, and, again, tcpdump shows the RAs
> > coming in.
> >
> > I've been fighting this for the past few weeks, and nothing I've found
> > when searching online has helped. I've also seen odd cases where I'm
> > able to get both a public IPv6 address and a default gateway, but only
> > on rare occasions.
> >
> > Does anybody have any suggestions?
> >
> > Thank you,
> >
> > - Neil
