RELIABILITY: No Graceful Shutdown

[devfire]

Severity: MEDIUM

Location

src/app.rs:94-100

Issue

Application just exits when stdin closes, with no cleanup:

tokio::select! {
    _ = udp_intake_handle => debug!("UDP intake task completed unexpectedly."),
    _ = display_handle => debug!("Display task completed unexpectedly."),
    _ = command_handle => debug!("Command task completed unexpectedly."),
    _ = message_handle => debug!("Message task completed unexpectedly."),
    _ = stdin_handle => debug!("Stdin task complete. Shutting down."),
}
// Just exits - no cleanup!

Problems

1. No state persistence: Groups and keys lost
2. No actor cleanup: Actors don't get shutdown signal
3. Network not closed: UDP socket left open
4. No graceful exit: Just terminates
5. No goodbye messages: Other users don't know we left

What Should Happen

Phase 1: Signal shutdown

// Create shutdown signal
let (shutdown_tx, shutdown_rx) = tokio::sync::broadcast::channel(1);

// On exit
debug!("Initiating graceful shutdown...");
shutdown_tx.send(()).ok();

Phase 2: Save state

// Persist groups and key packages
state_actor.ask(SaveStateMsg).await?;

Phase 3: Send leave messages

// Notify other users we're leaving
for group_name in groups {
    send_leave_message(group_name).await?;
}

Phase 4: Clean up resources

// Close network
network_manager.shutdown().await?;

// Stop actors
state_actor.stop().await?;
identity_actor.stop().await?;

Phase 5: Wait for completion

// Give tasks time to finish
tokio::time::timeout(
    Duration::from_secs(5),
    join_all(handles)
).await?;

Full Example

pub async fn run(&mut self) -> Result<()> {
    // ... spawn tasks ...
    
    let (shutdown_tx, mut shutdown_rx) = broadcast::channel(1);
    
    tokio::select! {
        _ = stdin_handle => {
            info!("User initiated shutdown");
            shutdown_tx.send(()).ok();
        }
        _ = signal::ctrl_c() => {
            info!("Received CTRL-C");
            shutdown_tx.send(()).ok();
        }
    }
    
    // Graceful shutdown sequence
    self.shutdown(&state_actor, &network_manager).await?;
    
    Ok(())
}

Benefits

• No data loss
• Clean resource cleanup
• Better user experience
• Proper multi-user behavior
• Can recover from crashes

Labels

reliability, enhancement, lifecycle