I think this has to do with how we are modeling spaces and dual-stack support. My guess is that when defining the LXD container we ask MAAS for an IP address in the requested space, but the space only defines an IPv4 subnet, so that is all we assign the container.
Even if we were assigning multiple addresses, this needs more discussion because of interactions with charms. When a given application has access to 2 addresses, what address do we advertise for other applications to interact with it. Do we give all addresses, a different address, etc. Right now our address field is a single string, so we don't have a way to expose it as a list of addresses.
I think this has to do with how we are modeling spaces and dual-stack support. My guess is that when defining the LXD container we ask MAAS for an IP address in the requested space, but the space only defines an IPv4 subnet, so that is all we assign the container.
Even if we were assigning multiple addresses, this needs more discussion because of interactions with charms. When a given application has access to 2 addresses, what address do we advertise for other applications to interact with it. Do we give all addresses, a different address, etc. Right now our address field is a single string, so we don't have a way to expose it as a list of addresses.