dhcpcd-discuss

On 18/06/2019 15:19, Breno Leitao wrote:
>    eth0: Router Advertisement from fe80::7270:8bff:fe48:3bbf
>    Address:  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>    Prefix:  26 20 1 d c0 82 24 14 0 0 0 0 0 0 f 4d
>    Prefix len = 128
>    ipv6_newaddr: Invalid argument

This tells me that a prefix was announced with a 128 length AND an 
instruction to make an autoconf address from it.
This is not possible because 128 is the maximum prefix length which 
means we can't add anything to it, hence invalid.

>    eth0: Router Advertisement from fe80::200:5eff:fe00:201
>    Address:  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
>    Prefix:  26 20 1 d c0 82 24 14 0 0 0 0 0 0 0 0
>    Prefix len = 64
>    eth0: adding address 2620:10d:c082:2414:aa1e:84ff:fef6:a07a/64
>    eth0: pltime 120 seconds, vltime 120 seconds

....

> That said, I think the problem is on network side, and dhcpcd just
> complained about a "invalid" RA.

The problem is definitely on the network side.
dhcpcd is receiving an RA from two routers. This is unusual for SOHO 
networks. When two routers are in play, they often log errors themselves 
if any other router broadcasts different information from what they are 
(at least NetBSD's rtadvd does).

The first RA received is in error, the second RA received looks valid.
fe80::7270:8bff:fe48:3bbf is a hardware derived LL address from, the 
first router.
72:70:8b:48:3b:bf is the hardware address of the node sending the faulty 
RA. However, that in turn looks like a virtual hardware address.
Good luck!

Roy