The persistent NetworkManager.Singleton.ShutdownInProgress state occurs because the internal state machine of the NetworkManager becomes desynchronized during concurrent shutdown requests. To fix this, your script must ensure the shutdown process is atomic and fully completed before resetting your state.

1. Create a private boolean flag named shuttingDown to act as a gatekeeper.
2. Wrap the NetworkManager.Singleton.Shutdown() call within a Coroutine to handle the asynchronous cleanup process.
3. Implement a loop that yields until the NetworkManager has cleared its active status and the ShutdownInProgress flag returns to false.
4. Reset shuttingDown only after the loop exits, ensuring the system is ready for a fresh connection.

This approach prevents redundant calls from OnClientStopped or other event listeners from interfering with the primary shutdown sequence.

Additional Tips:

• Always check if NetworkManager.Singleton is null before accessing members during a shutdown, as the manager might be destroyed if the scene is changing.
• If using the Relay service, ensure you also clean up any active allocations or join codes in your connection manager alongside the shuttingDown logic.
• Avoid calling Shutdown() inside OnDisable without checking the shuttingDown flag, as this can lead to race conditions during application exit.