feat: add retry logic for Envoy service discovery

Add 10-minute retry mechanism when getting Envoy service name to handle
cases where the service is not immediately available after deployment.

- Retry every 5 seconds for up to 10 minutes
- Show elapsed time in verbose mode
- Respect context cancellation

Signed-off-by: bitliu <[email protected]>

Author: Xunzhuo

e2e/profiles/ai-gateway/testcases.go

@@ -62,10 +62,34 @@ func testChatCompletionsRequest(ctx context.Context, client *kubernetes.Clientse
 		fmt.Println("[Test] Testing chat completions endpoint")
 	}
 
-	// Get the Envoy service name
-	envoyService, err := getEnvoyServiceName(ctx, opts.Verbose)
-	if err != nil {
-		return fmt.Errorf("failed to get Envoy service name: %w", err)
+	// Get the Envoy service name with retry
+	var envoyService string
+	var err error
+	retryTimeout := 10 * time.Minute
+	retryInterval := 5 * time.Second
+	startTime := time.Now()
+
+	for {
+		envoyService, err = getEnvoyServiceName(ctx, opts.Verbose)
+		if err == nil {
+			break
+		}
+
+		if time.Since(startTime) >= retryTimeout {
+			return fmt.Errorf("failed to get Envoy service name after %v: %w", retryTimeout, err)
+		}
+
+		if opts.Verbose {
+			fmt.Printf("[Test] Envoy service not found, retrying in %v... (elapsed: %v)\n",
+				retryInterval, time.Since(startTime).Round(time.Second))
+		}
+
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		case <-time.After(retryInterval):
+			// Continue retry
+		}
 	}
 
 	if opts.Verbose {