upgrade watchdog service to scheduled task (#1951)

cmd/launcher/launcher.go

@@ -302,7 +302,7 @@ func runLauncher(ctx context.Context, cancel func(), multiSlogger, systemMultiSl
 	go checkpointer.Once(ctx)
 	runGroup.Add("logcheckpoint", checkpointer.Run, checkpointer.Interrupt)
 
-	watchdogController, err := watchdog.NewController(ctx, k)
+	watchdogController, err := watchdog.NewController(ctx, k, opts.ConfigFilePath)
 	if err != nil { // log any issues here but move on, watchdog is not critical path
 		slogger.Log(ctx, slog.LevelError,
 			"could not init watchdog controller",

cmd/launcher/main.go

@@ -191,7 +191,7 @@ func runSubcommands(systemMultiSlogger *multislogger.MultiSlogger) error {
 	case "secure-enclave":
 		run = runSecureEnclave
 	case "watchdog": // note: this is currently only implemented for windows
-		run = watchdog.RunWatchdogService
+		run = watchdog.RunWatchdogTask
 	default:
 		return fmt.Errorf("unknown subcommand %s", os.Args[1])
 	}

docs/architecture/launcher_watchdog.md

@@ -3,31 +3,30 @@
 Note that for the initial implementation, this service is windows only. It is intentionally designed to give room for alternate OS implementations if needed in the future.
 Most of the relevant code can be found in [ee/watchdog](../../ee/watchdog/)
 
-Here is a basic sequence diagram displaying the enable path for the windows watchdog service. The `launcher_watchdog_enabled` control flag will trigger the initial configuration and installation, and removal of the flag will trigger removal of the service.
+Here is a basic sequence diagram displaying the enable path for the windows watchdog task. The `launcher_watchdog_enabled` control flag will trigger the initial configuration and installation, and removal of the flag will trigger removal of the task.
+
+You can alternatively install or remove the task for testing/troubleshooting using the `--install-task` and `--remove-task` options for the watchdog subcommand. Note this is intended for developer convenience or emergency usage - the `launcher_watchdog_enabled` flag sent from cloud will eventually override any manual actions.
 
 ```mermaid
 sequenceDiagram
     participant LauncherKolideK2Svc
     Note right of LauncherKolideK2Svc: ./launcher.exe svc ...
-    create participant WindowsServiceManager
-    LauncherKolideK2Svc->>WindowsServiceManager: if launcher_watchdog_enabled
-    create participant LauncherKolideWatchdogSvc
-    WindowsServiceManager->>LauncherKolideWatchdogSvc: have we installed the watchdog?
-    Note left of LauncherKolideWatchdogSvc: ./launcher.exe watchdog
+    participant WindowsSchedulerService
+    participant LauncherKolideK2WatchdogTask
+    Note right of LauncherKolideK2WatchdogTask: ./launcher.exe watchdog
 
-    alt yes the service already exists
-        LauncherKolideK2Svc->>LauncherKolideWatchdogSvc: Restart to ensure latest
-    else no the service does not exist
-        LauncherKolideK2Svc->>WindowsServiceManager: 1 - create, configure, etc
-        LauncherKolideK2Svc->>LauncherKolideWatchdogSvc: 2 - Start
-        activate LauncherKolideWatchdogSvc
+    alt launcher_watchdog_enabled
+        LauncherKolideK2Svc->>WindowsSchedulerService: create, configure, and install watchdog task
+        activate LauncherKolideK2WatchdogTask
+    else flag is not enabled
+        LauncherKolideK2Svc->>WindowsSchedulerService: remove watchdog task
     end
 
-    loop every n minutes
-        LauncherKolideWatchdogSvc->>WindowsServiceManager: Query LauncherKolideK2Svc status
-        LauncherKolideWatchdogSvc->>LauncherKolideK2Svc: Start if Stopped
+    loop every 30 minutes, or 1 minute after wake event
+        WindowsSchedulerService->>LauncherKolideK2WatchdogTask: triggers scheduled task
+        LauncherKolideK2WatchdogTask->>LauncherKolideK2Svc: performs healthcheck, restarts if stopped
     end
 ```
 
 The restart functionality is currently limited to detecting a stopped state, but the idea here is to lay out the foundation for more advanced healthchecking.
-The watchdog service itself runs as a separate invocation of launcher, writing all logs to sqlite. The main invocation of launcher runs a watchdog controller, which responds to the `launcher_watchdog_enabled` flag, and publishes all sqlite logs to debug.json.
+The watchdog task itself runs as a launcher subcommand, performing any required checks/actions and writing all logs to sqlite before exiting. The main invocation of launcher runs a watchdog controller, which responds to the `launcher_watchdog_enabled` flag, and publishes all sqlite logs to debug.json.

ee/powereventwatcher/power_event_watcher_windows.go

@@ -66,9 +66,9 @@ type (
 
 const (
 	eventIdEnteringModernStandby = 506
-	eventIdExitingModernStandby  = 507
+	EventIdExitingModernStandby  = 507
 	eventIdEnteringSleep         = 42
-	eventIdResumedFromSleep      = 107
+	EventIdResumedFromSleep      = 107
 
 	operationSuccessfulMsg = "The operation completed successfully."
 )
@@ -108,7 +108,7 @@ func (ims *InMemorySleepStateUpdater) OnPowerEvent(eventID int) error {
 	switch eventID {
 	case eventIdEnteringModernStandby, eventIdEnteringSleep:
 		ims.inModernStandby = true
-	case eventIdExitingModernStandby, eventIdResumedFromSleep:
+	case EventIdExitingModernStandby, EventIdResumedFromSleep:
 		ims.inModernStandby = false
 	default:
 		ims.slogger.Log(context.TODO(), slog.LevelWarn,
@@ -134,7 +134,7 @@ func (ks *knapsackSleepStateUpdater) OnPowerEvent(eventID int) error {
 				"err", err,
 			)
 		}
-	case eventIdExitingModernStandby, eventIdResumedFromSleep:
+	case EventIdExitingModernStandby, EventIdResumedFromSleep:
 		ks.slogger.Log(context.TODO(), slog.LevelDebug,
 			"system is waking",
 			"event_id", eventID,
@@ -187,9 +187,9 @@ func New(ctx context.Context, slogger *slog.Logger, pes powerEventSubscriber) (*
 
 	queryStr := fmt.Sprintf("*[System[Provider[@Name='Microsoft-Windows-Kernel-Power'] and (EventID=%d or EventID=%d or EventID=%d or EventID=%d)]]",
 		eventIdEnteringModernStandby,
-		eventIdExitingModernStandby,
+		EventIdExitingModernStandby,
 		eventIdEnteringSleep,
-		eventIdResumedFromSleep,
+		EventIdResumedFromSleep,
 	)
 	query, err := syscall.UTF16PtrFromString(queryStr)
 	if err != nil {

ee/uninstall/uninstall_windows.go

@@ -35,8 +35,8 @@ func disableAutoStart(ctx context.Context, k types.Knapsack) error {
 	}
 
 	// attempt to remove watchdog service in case it is installed to prevent startups later on
-	if err := watchdog.RemoveService(svcMgr); err != nil {
-		return fmt.Errorf("removing watchdog service, error may be expected if not enabled: %w", err)
+	if err := watchdog.RemoveWatchdogTask(k.Identifier()); err != nil {
+		return fmt.Errorf("removing watchdog task, error may be expected if not installed: %w", err)
 	}
 
 	return nil

ee/watchdog/controller_other.go

@@ -12,7 +12,7 @@ import (
 
 type WatchdogController struct{}
 
-func NewController(_ context.Context, _ types.Knapsack) (*WatchdogController, error) {
+func NewController(_ context.Context, _ types.Knapsack, _ string) (*WatchdogController, error) {
 	return nil, nil
 }