OKD-294: Migrate runtime from runc to crun on an upgrade for OKD

Centos stream 10 dopped the runc package. For now OKD users are
following the workaround to edit the MC to point to crun before the
upgrade, but this should help with the migration rather than having to
intervene manually.

This PR:
- checks for presence of the configmap
- if MCO is built for OKD, edit the MC so the runtime is crun.
- deletes the configmap

Author: Prashanth684

pkg/controller/container-runtime-config/container_runtime_config_controller.go

@@ -60,7 +60,9 @@ const (
 	// 5ms, 10ms, 20ms, 40ms, 80ms, 160ms, 320ms, 640ms, 1.3s, 2.6s, 5.1s, 10.2s, 20.4s, 41s, 82s
 	maxRetries = 15
 
-	builtInLabelKey = "machineconfiguration.openshift.io/mco-built-in"
+	builtInLabelKey    = "machineconfiguration.openshift.io/mco-built-in"
+	configMapName      = "crio-default-container-runtime"
+	forceSyncOnUpgrade = "force-sync-on-upgrade"
 )
 
 var (
@@ -79,6 +81,7 @@ type Controller struct {
 	templatesDir string
 
 	client        mcfgclientset.Interface
+	kubeClient    clientset.Interface
 	configClient  configclientset.Interface
 	eventRecorder record.EventRecorder
 
@@ -148,6 +151,7 @@ func New(
 	ctrl := &Controller{
 		templatesDir:  templatesDir,
 		client:        mcfgClient,
+		kubeClient:    kubeClient,
 		configClient:  configClient,
 		eventRecorder: ctrlcommon.NamespacedEventRecorder(eventBroadcaster.NewRecorder(scheme.Scheme, corev1.EventSource{Component: "machineconfigcontroller-containerruntimeconfigcontroller"})),
 		queue: workqueue.NewTypedRateLimitingQueueWithConfig(
@@ -213,6 +217,7 @@ func New(
 
 	ctrl.clusterVersionLister = clusterVersionInformer.Lister()
 	ctrl.clusterVersionListerSynced = clusterVersionInformer.Informer().HasSynced
+	ctrl.queue.Add(forceSyncOnUpgrade)
 
 	ctrl.fgHandler = fgHandler
 
@@ -596,6 +601,60 @@ func (ctrl *Controller) addAnnotation(cfg *mcfgv1.ContainerRuntimeConfig, annota
 	return annotationUpdateErr
 }
 
+// migrateRuncToCrun performs the upgrade migration from runc to crun as the default container runtime.
+// This function checks for the existence of the crio-default-container-runtime ConfigMap. If it exists,
+// it deletes the MachineConfigs for master and worker pools, then deletes the ConfigMap to prevent
+// re-running the migration.
+func (ctrl *Controller) migrateRuncToCrun() error {
+	// Check if the migration ConfigMap exists
+	_, err := ctrl.kubeClient.CoreV1().ConfigMaps(ctrlcommon.MCONamespace).Get(context.TODO(), configMapName, metav1.GetOptions{})
+	if errors.IsNotFound(err) {
+		// ConfigMap doesn't exist, no migration needed
+		return nil
+	}
+	if err != nil {
+		return fmt.Errorf("error checking for crio-default-container-runtime configmap: %w", err)
+	}
+
+	klog.Info("Found crio-default-container-runtime ConfigMap, starting migration from runc to crun")
+
+	// Get all MachineConfigPools
+	pools, err := ctrl.mcpLister.List(labels.Everything())
+	if err != nil {
+		return fmt.Errorf("error listing MachineConfigPools: %w", err)
+	}
+
+	// Only process master and worker pools for the migration
+	for _, pool := range pools {
+		if pool.Name != ctrlcommon.MachineConfigPoolMaster && pool.Name != ctrlcommon.MachineConfigPoolWorker {
+			continue
+		}
+
+		// Get the MachineConfig name for this pool
+		mcName := fmt.Sprintf("00-override-%s-generated-crio-default-container-runtime", pool.Name)
+
+		// Delete the existing MachineConfig
+		err := ctrl.client.MachineconfigurationV1().MachineConfigs().Delete(context.TODO(), mcName, metav1.DeleteOptions{})
+		if errors.IsNotFound(err) {
+			klog.Infof("MachineConfig %s not found, skipping migration for pool %s", mcName, pool.Name)
+			continue
+		}
+		if err != nil {
+			return fmt.Errorf("error deleting MachineConfig %s: %w", mcName, err)
+		}
+
+		klog.Infof("Successfully deleted MachineConfig %s", mcName)
+	}
+
+	// Delete the ConfigMap after successful migration
+	if err := ctrl.kubeClient.CoreV1().ConfigMaps(ctrlcommon.MCONamespace).Delete(context.TODO(), configMapName, metav1.DeleteOptions{}); err != nil && !errors.IsNotFound(err) {
+		return fmt.Errorf("error deleting crio-default-container-runtime configmap: %w", err)
+	}
+
+	klog.Info("Successfully completed migration from runc to crun and deleted migration ConfigMap")
+	return nil
+}
+
 // syncContainerRuntimeConfig will sync the ContainerRuntimeconfig with the given key.
 // This function is not meant to be invoked concurrently with the same key.
 // nolint: gocyclo
@@ -606,6 +665,17 @@ func (ctrl *Controller) syncContainerRuntimeConfig(key string) error {
 		klog.V(4).Infof("Finished syncing ContainerRuntimeconfig %q (%v)", key, time.Since(startTime))
 	}()
 
+	// OKD only: Run the migration function at the start of sync
+	if version.IsSCOS() {
+		if err := ctrl.migrateRuncToCrun(); err != nil {
+			return fmt.Errorf("Error during runc to crun migration: %w", err)
+		}
+	}
+
+	if key == forceSyncOnUpgrade {
+		return nil
+	}
+
 	_, name, err := cache.SplitMetaNamespaceKey(key)
 	if err != nil {
 		return err

pkg/controller/container-runtime-config/helpers.go

@@ -49,12 +49,13 @@ const (
 	policyConfigPath                       = "/etc/containers/policy.json"
 	// CRIODropInFilePathLogLevel is the path at which changes to the crio config for log-level
 	// will be dropped in this is exported so that we can use it in the e2e-tests
-	CRIODropInFilePathLogLevel       = "/etc/crio/crio.conf.d/01-ctrcfg-logLevel"
-	crioDropInFilePathPidsLimit      = "/etc/crio/crio.conf.d/01-ctrcfg-pidsLimit"
-	crioDropInFilePathLogSizeMax     = "/etc/crio/crio.conf.d/01-ctrcfg-logSizeMax"
-	CRIODropInFilePathDefaultRuntime = "/etc/crio/crio.conf.d/01-ctrcfg-defaultRuntime"
-	imagepolicyType                  = "sigstoreSigned"
-	sigstoreRegistriesConfigFilePath = "/etc/containers/registries.d/sigstore-registries.yaml"
+	CRIODropInFilePathLogLevel                    = "/etc/crio/crio.conf.d/01-ctrcfg-logLevel"
+	crioDropInFilePathPidsLimit                   = "/etc/crio/crio.conf.d/01-ctrcfg-pidsLimit"
+	crioDropInFilePathLogSizeMax                  = "/etc/crio/crio.conf.d/01-ctrcfg-logSizeMax"
+	CRIODropInFilePathDefaultRuntime              = "/etc/crio/crio.conf.d/01-ctrcfg-defaultRuntime"
+	CRIODropInFilePathDefaultContainerRuntimeCrun = "/etc/crio/crio.conf.d/01-mc-defaultContainerRuntimeCrun"
+	imagepolicyType                               = "sigstoreSigned"
+	sigstoreRegistriesConfigFilePath              = "/etc/containers/registries.d/sigstore-registries.yaml"
 )
 
 var (
@@ -1211,3 +1212,15 @@ func imagePolicyConfigFileList(namespaceJSONs map[string][]byte) []generatedConf
 	}
 	return namespacedPolicyConfigFileList
 }
+
+// createDefaultContainerRuntimeFile creates a TOML config file for crun as the default container runtime
+func createDefaultContainerRuntimeFile() []generatedConfigFile {
+	generatedConfigFileList := make([]generatedConfigFile, 0)
+	tomlConf := tomlConfigCRIODefaultRuntime{}
+	tomlConf.Crio.Runtime.DefaultRuntime = string(mcfgv1.ContainerRuntimeDefaultRuntimeCrun)
+	generatedConfigFileList, err := addTOMLgeneratedConfigFile(generatedConfigFileList, CRIODropInFilePathDefaultContainerRuntimeCrun, tomlConf)
+	if err != nil {
+		klog.V(2).Infof("error setting default-container-runtime to crio.conf.d: %v", err)
+	}
+	return generatedConfigFileList
+}