Can this website manage my cluster?
This website provides project information and guides. Deploy the management platform separately to connect sources, configure modules, inspect advice, and authorize actions.
Which model should I choose?
Start with a baseline and compare resource error, peak underestimation, and runtime over the same history and horizon. Complexity does not replace workload performance. Available forecasts can also contribute to consensus.
Can I use HPA or KEDA at the same time?
They may manage other workloads, but must not control the same target replicas. Stop the old controller and release ownership before switching.
Why are predictions or scores missing?
First check whether prediction is enabled and whether No prediction is selected. New modules start with prediction disabled; configured fallback defaults are intentional. Then check sources, template labels, units, history, model runs, and freshness. Actual scores require the target time to arrive and matching observations to be collected.
Do running Pods immediately receive updated advice?
Admission affects new Pods; running configurations change through creation or rolling updates. Replica scaling is a separate action performed by the Operator.
How do I pause automation?
Enable module bypass or disable the corresponding global gate, then return to Observe or Recommend. Review actual resources before restoring earlier settings; bypass does not undo previous changes.