fix: 等待 Runner facade 就绪

internal/runnerbootstrap/executor.go

@@ -79,6 +79,19 @@ func RunExecutor(ctx context.Context, config ExecutorConfig) error {
 	server := &http.Server{Handler: proxy.Handler, ReadHeaderTimeout: 10 * time.Second}
 	serverErrors := make(chan error, 1)
 	go func() { serverErrors <- server.Serve(listener) }()
+	readyContext, cancelReady := context.WithTimeout(ctx, 2*time.Minute)
+	readyErr := waitForFacade(readyContext, "http://"+listener.Addr().String())
+	cancelReady()
+	if readyErr != nil {
+		shutdownContext, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+		shutdownErr := server.Shutdown(shutdownContext)
+		cancel()
+		serverErr := <-serverErrors
+		if errors.Is(serverErr, http.ErrServerClosed) {
+			serverErr = nil
+		}
+		return errors.Join(readyErr, shutdownErr, serverErr)
+	}
 
 	command := exec.CommandContext(ctx, config.RunnerBinary, "daemon", "--once")
 	command.Dir = workDir
@@ -95,6 +108,30 @@ func RunExecutor(ctx context.Context, config ExecutorConfig) error {
 	return errors.Join(runnerErr, shutdownErr, serverErr)
 }
 
+func waitForFacade(ctx context.Context, endpoint string) error {
+	client := &http.Client{Timeout: 2 * time.Second}
+	ticker := time.NewTicker(250 * time.Millisecond)
+	defer ticker.Stop()
+	for {
+		request, err := http.NewRequestWithContext(ctx, http.MethodGet, endpoint, nil)
+		if err != nil {
+			return fmt.Errorf("create facade readiness request: %w", err)
+		}
+		response, requestErr := client.Do(request)
+		if requestErr == nil {
+			_ = response.Body.Close()
+			if response.StatusCode < http.StatusInternalServerError {
+				return nil
+			}
+		}
+		select {
+		case <-ctx.Done():
+			return fmt.Errorf("wait for runner facade: %w", ctx.Err())
+		case <-ticker.C:
+		}
+	}
+}
+
 // ExecutorConfigFromEnvironment reads the non-secret image configuration and
 // the assignment-scoped values injected by the backend. The Workload API
 // address follows SPIFFE_ENDPOINT_SOCKET through go-spiffe when not set here.

internal/runnerbootstrap/executor_test.go

@@ -0,0 +1,44 @@
+package runnerbootstrap
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"sync/atomic"
+	"testing"
+	"time"
+)
+
+func TestWaitForFacadeRetriesTransientGatewayFailure(t *testing.T) {
+	var requests atomic.Int32
+	server := httptest.NewServer(http.HandlerFunc(func(writer http.ResponseWriter, _ *http.Request) {
+		if requests.Add(1) < 3 {
+			writer.WriteHeader(http.StatusBadGateway)
+			return
+		}
+		writer.WriteHeader(http.StatusNotFound)
+	}))
+	defer server.Close()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+	defer cancel()
+	if err := waitForFacade(ctx, server.URL); err != nil {
+		t.Fatal(err)
+	}
+	if requests.Load() != 3 {
+		t.Fatalf("requests = %d", requests.Load())
+	}
+}
+
+func TestWaitForFacadeStopsWithContext(t *testing.T) {
+	server := httptest.NewServer(http.HandlerFunc(func(writer http.ResponseWriter, _ *http.Request) {
+		writer.WriteHeader(http.StatusServiceUnavailable)
+	}))
+	defer server.Close()
+
+	ctx, cancel := context.WithTimeout(context.Background(), 100*time.Millisecond)
+	defer cancel()
+	if err := waitForFacade(ctx, server.URL); err == nil {
+		t.Fatal("expected readiness timeout")
+	}
+}