From 39fbbbc2a6576ff213f62b0ac4612a2335b7a3a6 Mon Sep 17 00:00:00 2001
From: Nikolay Edigaryev <edigaryev@gmail.com>
Date: Wed, 16 Jul 2025 22:58:13 +0200
Subject: [PATCH] Disable Prometheus metrics by default (#331)

---
 internal/command/controller/run.go         |  9 ++++++++-
 internal/controller/controller.go          |  3 ++-
 internal/controller/option.go              |  6 ++++++
 internal/controller/scheduler/scheduler.go | 19 +++++++++++++------
 4 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/internal/command/controller/run.go b/internal/command/controller/run.go
index 482a282..38e46d9 100644
--- a/internal/command/controller/run.go
+++ b/internal/command/controller/run.go
@@ -31,6 +31,7 @@ var sshNoClientAuth bool
 var experimentalRPCV2 bool
 var noExperimentalRPCV2 bool
 var experimentalPingInterval time.Duration
+var deprecatedPrometheusMetrics bool
 
 func newRunCommand() *cobra.Command {
 	cmd := &cobra.Command{
@@ -69,10 +70,12 @@ func newRunCommand() *cobra.Command {
 	_ = cmd.Flags().MarkHidden("experimental-rpc-v2")
 	cmd.Flags().BoolVar(&noExperimentalRPCV2, "no-experimental-rpc-v2", false,
 		"disable experimental RPC v2 (https://github.com/cirruslabs/orchard/issues/235)")
-	cmd.PersistentFlags().DurationVar(&experimentalPingInterval, "experimental-ping-interval", 0,
+	cmd.Flags().DurationVar(&experimentalPingInterval, "experimental-ping-interval", 0,
 		"interval between WebSocket PING's sent by the controller to workers and clients, "+
 			"useful when facing intermediate load balancers/proxies that have timeouts "+
 			"smaller than the controller's default 30 second interval")
+	cmd.Flags().BoolVar(&deprecatedPrometheusMetrics, "deprecated-prometheus-metrics", false,
+		"enable Prometheus metrics, which will soon be deprecated in favor of OpenTelemetry")
 
 	return cmd
 }
@@ -180,6 +183,10 @@ func runController(cmd *cobra.Command, args []string) (err error) {
 		controllerOpts = append(controllerOpts, controller.WithPingInterval(experimentalPingInterval))
 	}
 
+	if deprecatedPrometheusMetrics {
+		controllerOpts = append(controllerOpts, controller.WithPrometheusMetrics())
+	}
+
 	controllerInstance, err := controller.New(controllerOpts...)
 	if err != nil {
 		return err
diff --git a/internal/controller/controller.go b/internal/controller/controller.go
index 59538f7..a32d444 100644
--- a/internal/controller/controller.go
+++ b/internal/controller/controller.go
@@ -61,6 +61,7 @@ type Controller struct {
 	maxWorkersPerLicense uint
 	experimentalRPCV2    bool
 	pingInterval         time.Duration
+	prometheusMetrics    bool
 
 	sshListenAddr   string
 	sshSigner       ssh.Signer
@@ -122,7 +123,7 @@ func New(opts ...Option) (*Controller, error) {
 
 	// Instantiate the scheduler
 	controller.scheduler, err = scheduler.NewScheduler(store, controller.workerNotifier,
-		controller.workerOfflineTimeout, controller.logger)
+		controller.workerOfflineTimeout, controller.prometheusMetrics, controller.logger)
 	if err != nil {
 		return nil, err
 	}
diff --git a/internal/controller/option.go b/internal/controller/option.go
index 23687a9..d2e8941 100644
--- a/internal/controller/option.go
+++ b/internal/controller/option.go
@@ -65,6 +65,12 @@ func WithPingInterval(pingInterval time.Duration) Option {
 	}
 }
 
+func WithPrometheusMetrics() Option {
+	return func(controller *Controller) {
+		controller.prometheusMetrics = true
+	}
+}
+
 func WithLogger(logger *zap.Logger) Option {
 	return func(controller *Controller) {
 		controller.logger = logger.Sugar()
diff --git a/internal/controller/scheduler/scheduler.go b/internal/controller/scheduler/scheduler.go
index 85995d6..8fa3560 100644
--- a/internal/controller/scheduler/scheduler.go
+++ b/internal/controller/scheduler/scheduler.go
@@ -49,6 +49,7 @@ type Scheduler struct {
 	workerOfflineTimeout time.Duration
 	logger               *zap.SugaredLogger
 	schedulingRequested  chan bool
+	prometheusMetrics    bool
 
 	schedulingTimeHistogram metric.Float64Histogram
 }
@@ -57,6 +58,7 @@ func NewScheduler(
 	store storepkg.Store,
 	notifier *notifier.Notifier,
 	workerOfflineTimeout time.Duration,
+	prometheusMetrics bool,
 	logger *zap.SugaredLogger,
 ) (*Scheduler, error) {
 	scheduler := &Scheduler{
@@ -65,6 +67,7 @@ func NewScheduler(
 		workerOfflineTimeout: workerOfflineTimeout,
 		logger:               logger,
 		schedulingRequested:  make(chan bool, 1),
+		prometheusMetrics:    prometheusMetrics,
 	}
 
 	// Metrics
@@ -104,7 +107,9 @@ func (scheduler *Scheduler) Run() {
 
 		if err != nil {
 			scheduler.logger.Errorf("Failed to schedule VMs: %v", err)
-		} else {
+		}
+
+		if scheduler.prometheusMetrics {
 			schedulerLoopIterationStat.Inc()
 		}
 	}
@@ -395,12 +400,14 @@ func (scheduler *Scheduler) healthCheckingLoopIteration() error {
 		}
 
 		// Update metrics
-		workers, err := txn.ListWorkers()
-		if err != nil {
-			return err
-		}
+		if scheduler.prometheusMetrics {
+			workers, err := txn.ListWorkers()
+			if err != nil {
+				return err
+			}
 
-		scheduler.reportStats(workers, vms)
+			scheduler.reportStats(workers, vms)
+		}
 
 		return nil
 	}); err != nil {