If I understand things correctly, MAAS will prefer a layer-2 connection over a routable connection from a rack controller to a BMC for power control, but all rack controllers with valid routes to the BMC are treated equally. In my situation, I have very few layer-2 connections to rack controllers, and rely on the routable connections. However, my rack controllers are spread out among 4 data centers, and doing a power_query_all against a given BMC can timeout, especially when using the Redfish driver. Is there, or could there be, a way to prefer or designate a rack controller (say, within the same data center), to control power?
Hi, what sort of connectivity do you have between your DCs, and what kind of latency RTT between them?
It seems strange that, with a decent network connection that you would experience timeouts.
It may very well be because the hardware vendor’s implementation of Redfish can’t handle concurrent power queries from many sources. If I isolate down to one or two rack controllers, the response is pretty snappy.