Fix critical/severe issues in init, startup and shutdown

- auth: do not abort the server when authentication is disabled. auth.Init
  is now always called; with disable-authentication it sets up an ephemeral
  session store (SESSION_KEY not required) and registers no authenticators,
  so the unconditional auth.GetAuthInstance() callers (server init,
  api.New()) always get a valid instance.
- main: run the graceful-shutdown sequence on the startup-error path. runServer
  derives a cancelable context and, on a server-start failure, cancels it and
  waits so the metricstore final checkpoint / WAL rotation, archiver flush and
  taskmanager shutdown actually run before exit.
- server: log the :80 HTTP->HTTPS redirect listener error instead of dropping it.
- archiver: guard Shutdown against being called when Start never ran
  (avoids close(nil) panic / blocking on a nil workerDone).
- nats API: stop worker goroutines on shutdown via a stop channel + idempotent
  Shutdown(); workers and subscription callbacks select on stop and the
  channels are never closed, so no send-on-closed-channel can occur. Wired
  into Server.Shutdown after the NATS client is closed.
- metricstore: make Shutdown idempotent (nil shutdownFunc, early return) and
  release shutdownFuncMu before the checkpoint write.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Entire-Checkpoint: 3c179f9caa8f

cmd/cc-backend/main.go

@@ -172,14 +172,20 @@ func handleUserCommands() error {
 		return fmt.Errorf("--add-user and --del-user can only be used if authentication is enabled")
 	}
 
-	if !config.Keys.DisableAuthentication {
-		if cfg := ccconf.GetPackageConfig("auth"); cfg != nil {
-			auth.Init(&cfg)
-		} else {
-			cclog.Warn("Authentication disabled due to missing configuration")
-			auth.Init(nil)
+	// Always initialize the auth subsystem so the HTTP server and REST API have a
+	// valid (non-nil) auth instance, even when authentication is disabled. With
+	// authentication disabled, Init only sets up an ephemeral session store and
+	// registers no authenticators (see auth.Init).
+	if cfg := ccconf.GetPackageConfig("auth"); cfg != nil {
+		auth.Init(&cfg)
+	} else {
+		if !config.Keys.DisableAuthentication {
+			cclog.Warn("Authentication enabled but no auth configuration found")
 		}
+		auth.Init(nil)
+	}
 
+	if !config.Keys.DisableAuthentication {
 		// Check for default security keys
 		checkDefaultSecurityKeys()
 
@@ -337,6 +343,12 @@ func initSubsystems() error {
 }
 
 func runServer(ctx context.Context) error {
+	// Derive a cancelable context so the startup-error path below can trigger the
+	// same graceful-shutdown sequence as a signal (via the signal handler that
+	// waits on ctx.Done()).
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+
 	var wg sync.WaitGroup
 
 	// Initialize metric store if configuration is provided
@@ -438,26 +450,32 @@ func runServer(ctx context.Context) error {
 	// Wait for either:
 	// 1. An error from server startup
 	// 2. Completion of all goroutines (normal shutdown or crash)
+	var runErr error
 	select {
-	case err := <-errChan:
+	case runErr = <-errChan:
 		// errChan will be closed when waitDone is closed, which happens
 		// when all goroutines complete (either from normal shutdown or error)
-		if err != nil {
-			return err
-		}
 	case <-time.After(100 * time.Millisecond):
 		// Give the server 100ms to start and report any immediate startup errors
 		// After that, just wait for normal shutdown completion
 		select {
-		case err := <-errChan:
-			if err != nil {
-				return err
-			}
+		case runErr = <-errChan:
 		case <-waitDone:
 			// Normal shutdown completed
 		}
 	}
 
+	if runErr != nil {
+		// A subsystem failed (e.g. the HTTP server could not bind). Trigger the
+		// graceful-shutdown path for the subsystems that were already started
+		// (metricstore checkpoint, archiver flush, taskmanager) by cancelling the
+		// context the signal handler waits on, then wait for it to finish so we
+		// don't exit before the final checkpoint is written.
+		cancel()
+		<-waitDone
+		return runErr
+	}
+
 	cclog.Print("Graceful shutdown completed!")
 	return nil
 }

cmd/cc-backend/server.go

@@ -415,7 +415,9 @@ func (s *Server) Start(ctx context.Context) error {
 
 	if !strings.HasSuffix(config.Keys.Addr, ":80") && config.Keys.RedirectHTTPTo != "" {
 		go func() {
-			http.ListenAndServe(":80", http.RedirectHandler(config.Keys.RedirectHTTPTo, http.StatusMovedPermanently))
+			if err := http.ListenAndServe(":80", http.RedirectHandler(config.Keys.RedirectHTTPTo, http.StatusMovedPermanently)); err != nil {
+				cclog.Errorf("HTTP-to-HTTPS redirect listener on :80 failed: %v", err)
+			}
 		}()
 	}
 
@@ -460,6 +462,11 @@ func (s *Server) Shutdown(ctx context.Context) {
 	if nc != nil {
 		nc.Close()
 	}
+	// Stop the NATS API worker goroutines after the client is closed (no more
+	// subscription callbacks can enqueue once the connection is down).
+	if s.natsAPIHandle != nil {
+		s.natsAPIHandle.Shutdown()
+	}
 	cclog.Infof("Shutdown: NATS closed (%v)", time.Since(natsStart))
 
 	httpStart := time.Now()