|
|
Post by iamsarki on Sept 8, 2023 12:42:55 GMT
Hi, I am still new to Redfish and forgive me if I ask something obvious. I am implementing Redfish on a device that acts as a supervisor of other devices. That is, it has other devices (that don't support Redfish) connected to it in various ways and it records their information and data. The idea was to implement Redfish only in the main one and expose some variables of the other devices in Redfish. My first approach was to create a 'manager' (AuxiliaryController) for the main device, and then a 'manager' (ManagementController) for every connected device and in them expose the variables in the OEM part since they could be anything. Does this implementation comply with the Redfish guidelines?
|
|
|
|
Post by mraineri on Sept 8, 2023 14:09:13 GMT
That's certainly one way to approach things and is allowed in the standard. My one suggestion with that is it sounds like your "main device" is the ManagementController, and your "other devices" map to AuxiliaryController; people tend to view auxiliary controllers as smaller with regards to their domain in a system.
It might be worth at least evaluating if you really need to show those "other devices" as their own Manager resources, or if they're simply conduits to extract data that goes to other places in the Redfish model. For example, you could technically model a microcontroller in a power supply as its own Manager resource, but this is quite extreme and doesn't really benefit the end user; the main surface for data in this case are the PowerSupply and PowerSupplyMetrics resources.
|
|
