RSDK-11248 Disable restart checking logic with new agents and update restart_status endpoint

robot/web/web.go

@@ -971,14 +971,23 @@ func (svc *webService) Stats() any {
 // RestartStatusResponse is the JSON response of the `restart_status` HTTP
 // endpoint.
 type RestartStatusResponse struct {
-	// RestartAllowed represents whether this instance of the viam-server can be
+	// RestartAllowed represents whether this instance of the viamserver can be
 	// safely restarted.
 	RestartAllowed bool `json:"restart_allowed"`
+	// DoesNotHandleNeedsRestart represents whether this instance of the viamserver does
+	// not check for the need to restart against app itself and, thus, needs agent to do so.
+	// Newer versions of viamserver (>= v0.9x.0) will report true for this value, while
+	// older versions won't report it at all, and agent should let viamserver handle
+	// NeedsRestart logic.
+	DoesNotHandleNeedsRestart bool `json:"does_not_handle_needs_restart,omitempty"`
 }
 
 // Handles the `/restart_status` endpoint.
 func (svc *webService) handleRestartStatus(w http.ResponseWriter, r *http.Request) {
-	response := RestartStatusResponse{RestartAllowed: svc.r.RestartAllowed()}
+	response := RestartStatusResponse{
+		RestartAllowed:            svc.r.RestartAllowed(),
+		DoesNotHandleNeedsRestart: true,
+	}
 
 	w.Header().Set("Content-Type", "application/json")
 	// Only log errors from encoding here. A failure to encode should never

web/cmd/droid/droid.go

@@ -19,7 +19,11 @@ var logger = logging.NewDebugLogger("droid-entrypoint")
 
 // DroidStopHook used by android harness to stop the RDK.
 func DroidStopHook() { //nolint:revive
-	server.ForceRestart = true
+	// NOTE(benjirewis): In RSDK-11248, we removed the restart checking logic from
+	// viam-server and put it in viam-agent. This method used to set a flag used by restart
+	// checking logic, but since that no longer exists and we have no users of this droid
+	// code, we no longer support this method.
+	logger.Error("DroidStopHook is no longer supported")
 }
 
 // MainEntry is called by our android app to start the RDK.

web/server/entrypoint.go

@@ -444,15 +444,11 @@ func (s *robotServer) serveWeb(ctx context.Context, cfg *config.Config) (err err
 	defer func() { <-forceShutdown }()
 
 	var (
-		theRobot                  robot.LocalRobot
-		theRobotLock              sync.Mutex
-		cloudRestartCheckerActive chan struct{}
+		theRobot     robot.LocalRobot
+		theRobotLock sync.Mutex
 	)
 	rpcDialer := rpc.NewCachedDialer()
 	defer func() {
-		if cloudRestartCheckerActive != nil {
-			<-cloudRestartCheckerActive
-		}
 		err = multierr.Combine(err, rpcDialer.Close())
 	}()
 	defer cancel()
@@ -529,33 +525,6 @@ func (s *robotServer) serveWeb(ctx context.Context, cfg *config.Config) (err err
 	// This functionality is tested in `TestLogPropagation` in `local_robot_test.go`.
 	config.UpdateLoggerRegistryFromConfig(s.registry, fullProcessedConfig, s.logger)
 
-	if fullProcessedConfig.Cloud != nil {
-		cloudRestartCheckerActive = make(chan struct{})
-		utils.PanicCapturingGo(func() {
-			defer close(cloudRestartCheckerActive)
-			restartCheck := newRestartChecker(cfg.Cloud, s.logger, s.conn)
-			restartInterval := defaultNeedsRestartCheckInterval
-
-			for {
-				if !utils.SelectContextOrWait(ctx, restartInterval) {
-					return
-				}
-
-				mustRestart, newRestartInterval, err := restartCheck.needsRestart(ctx)
-				if err != nil {
-					s.logger.Infow("failed to check restart", "error", err)
-					continue
-				}
-
-				restartInterval = newRestartInterval
-
-				if mustRestart {
-					logStackTraceAndCancel(cancel, s.logger)
-				}
-			}
-		})
-	}
-
 	robotOptions := createRobotOptions()
 	if s.args.RevealSensitiveConfigDiffs {
 		robotOptions = append(robotOptions, robotimpl.WithRevealSensitiveConfigDiffs())