fix: scheduler race — track loop goroutines in WaitGroup

Root cause: WaitForCompletion only waited for work goroutines (wg),
but the 5-6 loop goroutines (renewalCheckLoop, jobProcessorLoop, etc.)
were not tracked. After cancel() + WaitForCompletion(), loop goroutines
could still be alive accessing scheduler/mock fields when the next test
started, triggering the race detector.

Fix:
- Start() now adds loop goroutines to wg, so WaitForCompletion blocks
  until both work items AND loops have fully exited
- Removed untracked 100ms timer goroutine for startedChan — now closed
  immediately after launching loops
- Timeout test updated: uses blockCh (ignores context) instead of
  slowDelay (respects context) so it reliably triggers the timeout path

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

internal/scheduler/scheduler.go

@@ -126,21 +126,25 @@ func (s *Scheduler) Start(ctx context.Context) <-chan struct{} {
 	go func() {
 		s.logger.Info("scheduler starting")
 
-		// Signal that the scheduler has started all loops
+		// Track all loop goroutines in the WaitGroup so WaitForCompletion
-		go func() {
+		// blocks until they've fully exited (prevents test races).
-			<-time.After(100 * time.Millisecond)
+		loopCount := 5
-			close(startedChan)
-		}()
-
-		// Start all scheduler loops concurrently
-		go s.renewalCheckLoop(ctx)
-		go s.jobProcessorLoop(ctx)
-		go s.agentHealthCheckLoop(ctx)
-		go s.notificationProcessLoop(ctx)
-		go s.shortLivedExpiryCheckLoop(ctx)
 		if s.networkScanService != nil {
-			go s.networkScanLoop(ctx)
+			loopCount = 6
 		}
+		s.wg.Add(loopCount)
+
+		go func() { defer s.wg.Done(); s.renewalCheckLoop(ctx) }()
+		go func() { defer s.wg.Done(); s.jobProcessorLoop(ctx) }()
+		go func() { defer s.wg.Done(); s.agentHealthCheckLoop(ctx) }()
+		go func() { defer s.wg.Done(); s.notificationProcessLoop(ctx) }()
+		go func() { defer s.wg.Done(); s.shortLivedExpiryCheckLoop(ctx) }()
+		if s.networkScanService != nil {
+			go func() { defer s.wg.Done(); s.networkScanLoop(ctx) }()
+		}
+
+		// Signal that all loops are launched
+		close(startedChan)
 
 		// Wait for context cancellation
 		<-ctx.Done()

internal/scheduler/scheduler_test.go

@@ -16,14 +16,22 @@ type mockRenewalService struct {
 	callTimes   []time.Time
 	slowDelay   time.Duration
 	shouldError bool
+	blockCh     chan struct{} // if non-nil, blocks until closed (ignores context)
 }
 
 func (m *mockRenewalService) CheckExpiringCertificates(ctx context.Context) error {
 	m.mu.Lock()
 	m.callCount++
 	m.callTimes = append(m.callTimes, time.Now())
+	blockCh := m.blockCh
 	m.mu.Unlock()
 
+	// If blockCh is set, block until it's closed (ignores context — for timeout tests)
+	if blockCh != nil {
+		<-blockCh
+		return nil
+	}
+
 	if m.slowDelay > 0 {
 		select {
 		case <-time.After(m.slowDelay):
@@ -274,8 +282,9 @@ func TestWaitForCompletionTimeout(t *testing.T) {
 	// Use a channel-blocked mock that ignores context cancellation,
 	// ensuring work is still in-flight when WaitForCompletion is called.
 	blockCh := make(chan struct{})
-	renewalMock := &mockRenewalService{}
+	renewalMock := &mockRenewalService{
-	renewalMock.slowDelay = 0 // We override behavior below
+		blockCh: blockCh, // blocks until closed, ignores ctx
+	}
 
 	jobMock := &mockJobService{}
 	agentMock := &mockAgentService{}
@@ -290,31 +299,23 @@ func TestWaitForCompletionTimeout(t *testing.T) {
 	defer cancel()
 	defer close(blockCh) // Unblock the mock after test completes
 
-	// Override the renewal mock to block on a channel (ignores context cancel)
-	renewalMock.slowDelay = 30 * time.Second // Long enough to outlast the test
-
 	// Start scheduler
 	startedChan := sched.Start(ctx)
 	<-startedChan
 
-	// Let it run briefly so a job starts
+	// Let it run briefly so the initial job starts and blocks
-	time.Sleep(150 * time.Millisecond)
+	time.Sleep(50 * time.Millisecond)
 
-	// Stop scheduler — but the in-flight job won't finish (blocked)
+	// Stop scheduler — but the in-flight work goroutine won't finish (blocked on channel)
 	cancel()
 
-	// Wait with very short timeout (much shorter than the blocked job)
+	// Wait with very short timeout (work is stuck on blockCh)
 	start := time.Now()
 	err := sched.WaitForCompletion(200 * time.Millisecond)
 	elapsed := time.Since(start)
 
-	if err == nil {
-		t.Logf("WaitForCompletion completed in %v (job may have been cancelled by context)", elapsed)
-		t.Skip("flaky: job completed before timeout — context cancellation propagated faster than expected")
-	}
-
 	if err != ErrSchedulerShutdownTimeout {
-		t.Fatalf("expected ErrSchedulerShutdownTimeout, got %v", err)
+		t.Fatalf("expected ErrSchedulerShutdownTimeout, got %v (elapsed: %v)", err, elapsed)
 	}
 
 	t.Logf("WaitForCompletion correctly timed out after %v", elapsed)