Chore: Attempt to fix flaky tests that fail often in CI (#6004)

Attempts to fix some of these flaky tests that fail often in CI:

Author: wbreza

cli/azd/internal/appdetect/java_test.go

@@ -5,17 +5,26 @@ package appdetect
 
 import (
 	"context"
+	"fmt"
 	"log/slog"
 	"os"
 	osexec "os/exec"
 	"path/filepath"
+	"strings"
 	"testing"
+	"time"
 
 	"github.com/azure/azure-dev/cli/azd/pkg/exec"
 	"github.com/azure/azure-dev/cli/azd/pkg/tools/maven"
+	"github.com/sethvargo/go-retry"
 )
 
 func TestToMavenProject(t *testing.T) {
+	// Skip in short mode since this test requires network access to Maven Central
+	if testing.Short() {
+		t.Skip("Skipping Maven network-dependent test in short mode")
+	}
+
 	path, err := osexec.LookPath("java")
 	if err != nil {
 		t.Skip("Skip readMavenProject because java command doesn't exist.")
@@ -358,7 +367,12 @@ func TestToMavenProject(t *testing.T) {
 			testPom := tt.testPoms[0]
 			pomFilePath := filepath.Join(workingDir, testPom.pomFilePath)
 
-			mavenProject, err := readMavenProject(context.TODO(), maven.NewCli(exec.NewCommandRunner(nil)),
+			// Use a timeout context to prevent hanging on network issues
+			ctx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
+			defer cancel()
+
+			// Use retry logic for Maven operations due to potential network issues
+			mavenProject, err := readMavenProjectWithRetry(ctx, maven.NewCli(exec.NewCommandRunner(nil)),
 				pomFilePath)
 			if err != nil {
 				t.Fatalf("readMavenProject failed: %v", err)
@@ -377,6 +391,44 @@ func TestToMavenProject(t *testing.T) {
 	}
 }
 
+// readMavenProjectWithRetry wraps readMavenProject with retry logic to handle network issues
+func readMavenProjectWithRetry(ctx context.Context, mvnCli *maven.Cli, filePath string) (*mavenProject, error) {
+	var mavenProject *mavenProject
+	var lastErr error
+
+	err := retry.Do(
+		ctx,
+		retry.WithMaxRetries(3, retry.NewExponential(1*time.Second)),
+		func(ctx context.Context) error {
+			result, err := readMavenProject(ctx, mvnCli, filePath)
+			if err != nil {
+				// Check if error is likely network-related
+				errStr := strings.ToLower(err.Error())
+				if strings.Contains(errStr, "connection") ||
+					strings.Contains(errStr, "timeout") ||
+					strings.Contains(errStr, "network") ||
+					strings.Contains(errStr, "unknown host") ||
+					strings.Contains(errStr, "could not resolve") ||
+					strings.Contains(errStr, "transfer failed") {
+					lastErr = err
+					return retry.RetryableError(err)
+				}
+				// For non-network errors (parsing, etc.), fail immediately
+				return err
+			}
+			mavenProject = result
+			return nil
+		},
+	)
+
+	if err != nil && lastErr != nil {
+		// If we retried but still failed, include context about retries
+		return nil, fmt.Errorf("maven operation failed after retries due to network issues: %w", lastErr)
+	}
+
+	return mavenProject, err
+}
+
 type testPom struct {
 	pomFilePath      string
 	pomContentString string

cli/azd/test/functional/vs_server_test.go

@@ -51,18 +51,33 @@ func Test_CLI_VsServerExternalAuth(t *testing.T) {
 	err := cmd.Start()
 	require.NoError(t, err)
 
-	// Wait for the server to start
-	for i := 0; i < 5; i++ {
-		time.Sleep(300 * time.Millisecond)
+	// Wait for the server to start and output complete JSON
+	var svr contracts.VsServerResult
+	var outputData []byte
+	maxAttempts := 20 // Increased from 5 to give more time
+	for i := 0; i < maxAttempts; i++ {
+		time.Sleep(150 * time.Millisecond) // Reduced sleep time but more attempts
 		if stdout.Len() > 0 {
-			break
+			outputData = stdout.Bytes()
+			// Try to parse JSON - if it succeeds, we have complete output
+			if err := json.Unmarshal(outputData, &svr); err == nil {
+				break
+			}
+			// If we're on the last attempt and still can't parse, fail with helpful error
+			if i == maxAttempts-1 {
+				require.NoError(
+					t,
+					err,
+					"failed to parse JSON after %d attempts, output: %s",
+					maxAttempts,
+					string(outputData),
+				)
+			}
+		} else if i == maxAttempts-1 {
+			require.FailNow(t, "server did not produce any output after %d attempts", maxAttempts)
 		}
 	}
 
-	var svr contracts.VsServerResult
-	err = json.Unmarshal(stdout.Bytes(), &svr)
-	require.NoError(t, err, "value: %s", stdout.String())
-
 	ssConn, _, err := websocket.DefaultDialer.Dial(fmt.Sprintf("ws://127.0.0.1:%d/ServerService/v1.0", svr.Port), nil)
 	require.NoError(t, err)
 
@@ -235,18 +250,33 @@ func Test_CLI_VsServer(t *testing.T) {
 			err = cmd.Start()
 			require.NoError(t, err)
 
-			// Wait for the server to start
-			for i := 0; i < 5; i++ {
-				time.Sleep(300 * time.Millisecond)
+			// Wait for the server to start and output complete JSON
+			var svr contracts.VsServerResult
+			var outputData []byte
+			maxAttempts := 20 // Increased from 5 to give more time
+			for i := 0; i < maxAttempts; i++ {
+				time.Sleep(150 * time.Millisecond) // Reduced sleep time but more attempts
 				if stdout.Len() > 0 {
-					break
+					outputData = stdout.Bytes()
+					// Try to parse JSON - if it succeeds, we have complete output
+					if err := json.Unmarshal(outputData, &svr); err == nil {
+						break
+					}
+					// If we're on the last attempt and still can't parse, fail with helpful error
+					if i == maxAttempts-1 {
+						require.NoError(
+							t,
+							err,
+							"failed to parse JSON after %d attempts, output: %s",
+							maxAttempts,
+							string(outputData),
+						)
+					}
+				} else if i == maxAttempts-1 {
+					require.FailNow(t, "server did not produce any output after %d attempts", maxAttempts)
 				}
 			}
 
-			var svr contracts.VsServerResult
-			err = json.Unmarshal(stdout.Bytes(), &svr)
-			require.NoError(t, err, "value: %s", stdout.String())
-
 			/* #nosec G204 - Subprocess launched with a potential tainted input or cmd arguments false positive */
 			cmd = exec.CommandContext(context.Background(),
 				"dotnet", "test",

cli/azd/test/recording/recording.go

@@ -195,10 +195,14 @@ func Start(t *testing.T, opts ...Options) *Session {
 	}
 
 	transport := http.DefaultTransport.(*http.Transport).Clone()
-	vcr.SetRealTransport(&gzip2HttpRoundTripper{
+
+	// Wrap the transport chain with resilient retry logic and gzip handling
+	resilientTransport := NewResilientHttpTransport(&gzip2HttpRoundTripper{
 		transport: transport,
 	})
 
+	vcr.SetRealTransport(resilientTransport)
+
 	vcr.SetMatcher(func(r *http.Request, i cassette.Request) bool {
 		// Ignore query parameter 's=...' in containerappOperationResults
 		if strings.Contains(r.URL.Path, "/providers/Microsoft.App/") &&

cli/azd/test/recording/resilient_transport.go

@@ -0,0 +1,89 @@
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License.
+
+package recording
+
+import (
+	"context"
+	"net/http"
+	"strings"
+	"time"
+
+	"github.com/sethvargo/go-retry"
+)
+
+// networkErrorKeywords contains error message keywords that indicate network-related failures
+// that should be retried
+var networkErrorKeywords = []string{
+	"timeout",
+	"connection",
+	"network",
+	"dns",
+	"tls handshake",
+	"context deadline exceeded",
+	"request timeout",
+	"no such host",
+	"connection refused",
+	"connection reset",
+	"i/o timeout",
+	"network is unreachable",
+	"temporary failure",
+	"service unavailable",
+}
+
+// resilientHttpTransport wraps an HTTP transport with retry logic for network failures.
+// This makes the test recorder more robust to transient network issues without affecting recorded interactions.
+type resilientHttpTransport struct {
+	transport http.RoundTripper
+}
+
+// NewResilientHttpTransport creates a new resilient HTTP transport that wraps the provided transport
+func NewResilientHttpTransport(transport http.RoundTripper) *resilientHttpTransport {
+	return &resilientHttpTransport{
+		transport: transport,
+	}
+}
+
+// isNetworkError checks if an error message contains keywords indicating a network-related failure
+func isNetworkError(err error) bool {
+	if err == nil {
+		return false
+	}
+
+	errStr := strings.ToLower(err.Error())
+	for _, keyword := range networkErrorKeywords {
+		if strings.Contains(errStr, keyword) {
+			return true
+		}
+	}
+	return false
+}
+
+// RoundTrip implements http.RoundTripper with retry logic for network failures
+func (r *resilientHttpTransport) RoundTrip(req *http.Request) (*http.Response, error) {
+	var resp *http.Response
+	var err error
+
+	// Retry logic with exponential backoff for network failures
+	retryErr := retry.Do(
+		req.Context(),
+		retry.WithMaxRetries(3, retry.NewExponential(2*time.Second)),
+		func(ctx context.Context) error {
+			resp, err = r.transport.RoundTrip(req)
+			if err != nil {
+				// Check if error is likely network-related
+				if isNetworkError(err) {
+					return retry.RetryableError(err)
+				}
+				// For non-network errors, fail immediately
+				return err
+			}
+			return nil
+		},
+	)
+
+	if retryErr != nil {
+		return nil, retryErr
+	}
+	return resp, nil
+}