The ECS CPU monitoring result in the CloudMonitor is 0 or a negative value (the actual CPU usage is not 0), and other monitoring results are normal. This problem primarily happens to Windows servers. A general cause for it is the damage to the internal Windows performance counter.
You can run the typeperf \Processor(_Total)\% Processor Time command to check whether the counter works properly. If the result is Error: no valid counter, it indicates that the counter has failed. You can run the lodctr /r command to fix it.