LED icons change to yellow or red only when they are down. There is also a thing called "child status" that is represented by a little square in the right-bottom corner of the LED. This status shows the worst status of all nodes' interfaces (for instance).
This behavior is slightly modified in maps. The child status also covers Metrics. These are statistics with thresholds. If a value of a metric goes over a threshold, it influences the child status of node (only on the map).
You can set the thresholds in Network Atlas when you right click on a node on a map, select Properties, and then the Metrics tab. CPU Load is one of the monitored metrics. The thresholds are stored in the map, so each map may have different thresholds for different nodes.
Thanks for your response. My experience confirms what you have said. I have a node at 92% physical memory. On the "All Nodes" list, it shows green. On the Map, there is a yellow "thing" at the lower right of the Icon.
From a behavior standpoint, it would make sense if the node would change color if anything is wrong with the node. Right now, I have to look in several places to see if there is a problem.