Currently, RPC nodes operate passively and may not be aware of impending or active network upgrades. The risk is that nodes may continue serving requests during network halt.
We need to design a mechanism for RPC nodes to autonomously detect when a network upgrade is active or imminent and transition into a well-defined “network upgrade detected” state. This way clients like console can implement appropriate measures to handle this, such as offline mode.