Normal Background: There are customer scenarios where the edge device will not have direct wired access to a physical switch but the device still has the capability to run OpenShift and the customer has a need to actually extend their cloud out to the edge with consistency. In these cases there might need to be a way for at least Single Node OpenShift to be able to communicate to the internet via a Wifi/WAN/Satellite connection.
Description: Enable Single Node OpenShift to use Wifi/Wan/Satellite connection
Experience Today: When you install SNO on a device with Wifi adapter that is supported by RHEL the device will show up with lspci in RHCOS:
[core@84-8b-cd-40-55-86 ~]$ lspci | grep -i wireless01:00.0 Network controller: Intel Corporation Wireless-AC 9260 (rev 29)
However there is no tooling and/or maybe missing firmware which does not enable the device to show up as a usable interface via the ip command:
[core@84-8b-cd-40-55-86 ~]$ ip addr | grep -i wlp
[core@84-8b-cd-40-55-86 ~]$
Along with that there is no options to configure SSID and authentication credentials for the Wifi base to connect.
Considerations:
-How do we handle scenarios like a SNO device using Wifi on a train and the Wifi access points change. For example in the first 50 miles I am connected to SSID: spike1 but after mile 50 I need to now shift to connect to SSID spike2. Not only do we need multiple credentials but we also need to enable that transition capability within SNO to handle that my IP address might change.
-Use of Wifi could be important in relation to OCP appliance based deployment which could be used in a scenario where I want to ship OCP appliance and finish the installation at a site that only has Wifi/WAN/Satellite capabilities. Think of retailers and popup temporary stores.
-How do you pick which firmware related to the wireless adapters to also include in RHCOS? Including them all might increase image size of RHCOS so do we layer them on after or build them all in or specify in some configuration?
-Other considerations?