Having a weird situation going on with IPv6. Every machine on my network can successfully get an ipv6 address except for one. (My main server.) What's even weirder is that ipv6 was working on that machine up until probably about a week ago. Not sure what the problem is, and hoping someone here might be able to offer a pointer. Here's the scenario:
1) Wifi router reports an IPv6 address on both the WAN and the LAN sides. (In the 2607: range.) 2) Mac desktop behind the router successfully gets assigned an ipv6 address. 3) Linux laptop (running Arch) successfully gets assigned an ipv6 address (using NetworkManager). 4) Linux server (also running Arch) does not get assigned an ipv6 address.
The server is using netctl for networking, and its profile includes:
## For IPv6 autoconfiguration IP6=stateless
Yet for some reason, this machine doesn't get assigned an external ipv6. Rather it only gets assigned one in the link-local range. (I.e., in the fe80: range.)
I have to admit this is pretty puzzling. I can't for the life of me figure out why the server can't get an address. Particular since it was able to do so successfully probably about a week or so ago.
Anyone have an idea what could have happened here? Has there been some package update recently that might explain this? Or any pointers on how I might go about debugging this?
Thanks,
DR
Em dezembro 22, 2017 14:15 David Rosenstrauch escreveu:
- Wifi router reports an IPv6 address on both the WAN and the LAN
sides. (In the 2607: range.)
That's a global address range, for sure.
- Mac desktop behind the router successfully gets assigned an ipv6
address. 3) Linux laptop (running Arch) successfully gets assigned an ipv6 address (using NetworkManager). 4) Linux server (also running Arch) does not get assigned an ipv6 address.
The server is using netctl for networking, and its profile includes:
## For IPv6 autoconfiguration IP6=stateless
Try using stateful here.
Yet for some reason, this machine doesn't get assigned an external ipv6. Rather it only gets assigned one in the link-local range. (I.e., in the fe80: range.)
Link-local addresses are assigned to any interface when they are created, regardless of other IPv6 configuration.
I have to admit this is pretty puzzling. I can't for the life of me figure out why the server can't get an address. Particular since it was able to do so successfully probably about a week or so ago.
Anyone have an idea what could have happened here? Has there been some package update recently that might explain this? Or any pointers on how I might go about debugging this?
Well, stateless configuration, or SLAAC, relies heavily on ICMPv6 and NDP. A firewall can block these. If you try to use stateful configuration and it works, then you might be inadvertently be blocking the packets used on slaac.
Regards, Giancarlo Razzolini
arch-general@lists.archlinux.org