I don’t believe this is documented yet. I believe the approach will be: * All the OpenStack external endpoints will be behind an nginx ingress controller * i.e. all openstack service api endpoints and horizon * I know horizon is currently on a host port, but will be moved behind nginx eventually * Nginx is on ports 80 and 443 (http and https) * And NOTE uses the FQDN of the destination IP to route to the specific service * i.e. remote client MUST use FQDN * This also IMPLIES that the domain of the platform’s coredns server must be uniquely configured for the cloud, and the DNS server that you are using remotely must either delegate this domain to the platform’s coredns server or have these FQDNs configured on its own. Here’s the pic I drew: [cid:image001.png@01D4E85E.385C2600] Greg. From: Curtis <serverascode@gmail.com> Date: Friday, March 29, 2019 at 12:54 PM To: "starlingx-discuss@lists.starlingx.io" <starlingx-discuss@lists.starlingx.io> Subject: [Starlingx-discuss] Remote access to APIs, Horizon Hi All, Are there docs or what is the current thinking around remote access of APIs and Horizon for OpenStack deployments in STX? I looked around a bit but might of missed it if there are some. I see endpoints with cluster ips and a nodeport for Horizon on port 31000, after that I'm outta the loop. :) Thanks, Curtis
