Hi,
I’m trying to deploy OpenStack with JUJU and MAAS. I have several servers, each with four NICs. One is a management network (10.10.0.0/16), one is a MAAS-managed network with DHCP for provisioning servers (192.168.120.0/24), and two NICs (for redundancy) are on our external network (172.28.0.0/16), which is our lab network. In MAAS, mgmt, PXE, and LAB spaces have been created, and the fabrics mentioned above have been added to those spaces.
The provisioning network (192.168.120.0/24) is not routed to uplink switches, so the lab network can’t see it. The management network is routed and accessible from the lab network, but I can’t use it for running services due to security reasons.
So what’s left is the lab network itself. I have set the LAB space as the default space for Juju. This way, lab users can log in to the OpenStack dashboard in the future without further network configuration.
My first question is: do you see any issues with this setup?
The second, and main, issue is that the DHCP in the lab network also provides IPv6 to the machines. As a result, MAAS automatically assigns two subnets—IPv4 and IPv6—to the fabric. The problem now is that when JUJU deploys LXD containers, only IPv6 is assigned to them.
I read in another post that this was announced as a bug or something similar, but I can’t find any follow-up information.
First, I want to know if this is a problem. I read somewhere that OpenStack and IPv6 might cause issues. What is the solution to this?