Comment 11 for bug 1377005

And, this time I can confirm that during the deployment boot, the node successfully made its metadata requests through IPv6. *However* it doesn't seem to use the statically configured IPv6 address!

When I deploy a node with IPv4 disabled, on a network with router advertisements, the node seems to use an autoconfigured IPv6 address at one point in the middle of the procedure. First, during installation when IPv4 still works, cloud-init retrieves its preseed and curtin retrieves various metadata items, all through IPv4. Then, about 35 seconds later, ‘wget’ on the node does a POST to the metadata API from the autoconfigured IPv6 address. Then, some 42 seconds after that, the node retrieves more metadata items from its link-local IPv6 address.

So I see the node using IPv6, but I never get to see it using its configured IPv6 address.