fix: Setting of right exit code in case of detailed-exitcode (#4357)

* Detailed exit code update

* Added reset to success error

* Added unit test to track exit code

Author: denis256

cli/commands/run/run.go

@@ -485,6 +485,10 @@ func RunTerraformWithRetry(ctx context.Context, terragruntOptions *options.Terra
 				return err
 			} else {
 				terragruntOptions.Logger.Infof("Encountered an error eligible for retrying. Sleeping %v before retrying.\n", terragruntOptions.RetrySleepInterval)
+				// Reset the exit code to success so that we can retry the command
+				if exitCode := tf.DetailedExitCodeFromContext(ctx); exitCode != nil {
+					exitCode.ResetSuccess()
+				}
 				select {
 				case <-time.After(terragruntOptions.RetrySleepInterval):
 					// try again

test/integration_test.go

@@ -233,6 +233,32 @@ func TestDetailedExitCodeChangesPresentAll(t *testing.T) {
 	assert.Equal(t, 2, exitCode.Get())
 }
 
+func TestDetailedExitCodeChangesUnit(t *testing.T) {
+	t.Parallel()
+
+	testFixturePath := filepath.Join(testFixtureDetailedExitCode, "changes")
+
+	helpers.CleanupTerraformFolder(t, testFixturePath)
+	tmpEnvPath := helpers.CopyEnvironment(t, testFixturePath)
+	rootPath := util.JoinPath(tmpEnvPath, testFixturePath)
+	ctx := t.Context()
+
+	_, _, err := helpers.RunTerragruntCommandWithOutputWithContext(t, ctx, "terragrunt run --all --log-level trace --non-interactive --working-dir "+rootPath+" -- apply")
+	require.NoError(t, err)
+
+	// delete example.txt from rootPath/app1 to have changes in one unit
+	err = os.Remove(filepath.Join(rootPath, "app1", "example.txt"))
+	require.NoError(t, err)
+
+	// check that the exit code is 2 when there are changes in one unit
+	var exitCode tf.DetailedExitCode
+	ctx = tf.ContextWithDetailedExitCode(ctx, &exitCode)
+
+	_, _, err = helpers.RunTerragruntCommandWithOutputWithContext(t, ctx, "terragrunt run --all --log-level trace --non-interactive --working-dir "+rootPath+" -- plan -detailed-exitcode")
+	require.NoError(t, err)
+	assert.Equal(t, 2, exitCode.Get())
+}
+
 func TestDetailedExitCodeFailOnFirstRun(t *testing.T) {
 	t.Parallel()
 

tf/detailed_exitcode.go

@@ -4,27 +4,49 @@ import (
 	"sync"
 )
 
+const (
+	DetailedExitCodeSuccess = 0
+	DetailedExitCodeError   = 1
+)
+
 // DetailedExitCode is the TF detailed exit code. https://opentofu.org/docs/cli/commands/plan/
 type DetailedExitCode struct {
 	Code int
 	mu   sync.RWMutex
 }
 
-// Get returns exit code.
+// Get return exit code.
 func (coder *DetailedExitCode) Get() int {
 	coder.mu.RLock()
 	defer coder.mu.RUnlock()
 
 	return coder.Code
 }
 
+// ResetSuccess resets the exit code to success (0).
+func (coder *DetailedExitCode) ResetSuccess() {
+	coder.mu.Lock()
+	defer coder.mu.Unlock()
+
+	coder.Code = DetailedExitCodeSuccess
+}
+
 // Set updates the exit code following OpenTofu's exit code convention:
 // - 0 = Success
 // - 1 = Error
 // - 2 = Success with changes pending
+// The method only updates if:
+// - The current code is not 1 (error state)
+// - The new code is greater than current OR equals 1
 func (coder *DetailedExitCode) Set(newCode int) {
 	coder.mu.Lock()
 	defer coder.mu.Unlock()
 
-	coder.Code = newCode
+	if coder.Code == DetailedExitCodeError {
+		return
+	}
+
+	if coder.Code < newCode || newCode == DetailedExitCodeError {
+		coder.Code = newCode
+	}
 }

tf/run_cmd.go

@@ -74,10 +74,6 @@ func RunCommandWithOutput(ctx context.Context, opts *options.TerragruntOptions,
 		if code != 1 {
 			return output, nil
 		}
-	} else {
-		if exitCode := DetailedExitCodeFromContext(ctx); exitCode != nil {
-			exitCode.Set(0)
-		}
 	}
 
 	return output, err