Can a Docker container monitor the physical host’s CPU and RAM on macOS or Windows?

The practical cross-platform solution is to run a small metrics exporter on the host and have the container query it over the local network. For example, a Prometheus-compatible exporter can expose CPU and memory data from macOS or Windows, while your container reads the HTTP endpoint. It does mean running a host-side component, because the container cannot bypass the virtualization boundary by itself.

Increasing or changing Docker Desktop’s CPU and memory settings only changes the VM limits. It does not make the physical host’s metrics visible inside the container. Windows containers are a separate case, but a Linux-based container still cannot assume it can inspect the Windows host kernel directly.

You could theoretically expose host information through a mounted filesystem or another integration, but there is no macOS equivalent of Linux’s /proc that Docker can simply bind-mount. Any host filesystem data would use an OS-specific format and require custom parsing, so it would not provide a clean portable psutil solution.

Not directly. On macOS and Windows, Docker Desktop normally runs the Linux container inside a Linux VM. psutil and /proc can only see the kernel and resource limits of that VM, so the container will report the VM’s memory and CPU rather than the physical computer’s values. Native Linux is different because containers share the host Linux kernel.