Update ClusterLogForwarder status when ClusterLogging becomes valid

After a valid ClusterLogForwarder and an invalid ClusterLogging resource
are created, the ClusterLogForwarder's status will show that the CLF is
not available. When the ClusterLogging CR is then changes to a valid
configuration, the CLF's status should reflect that everything is o.k.
now.

Signed-off-by: Andreas Karis <ak.karis@gmail.com>

internal/controller/clusterlogging/clusterlogging_controller.go

@@ -108,10 +108,18 @@ func (r *ReconcileClusterLogging) Reconcile(ctx context.Context, request ctrl.Re
 		telemetry.Data.CLInfo.Set("managedStatus", constants.UnManagedStatus)
 		return ctrl.Result{}, nil
 	}
-	clf, err, _ := loader.FetchClusterLogForwarder(r.Client, request.NamespacedName.Namespace, request.NamespacedName.Name, false, func() loggingv1.ClusterLogging { return instance })
+	clf, err, cls := loader.FetchClusterLogForwarder(r.Client, request.NamespacedName.Namespace, request.NamespacedName.Name, false, func() loggingv1.ClusterLogging { return instance })
 	if err != nil && !errors.IsNotFound(err) {
 		return ctrl.Result{}, err
 	}
+	// Make sure that the ClusterForwarder status is updated as well, as we might migrate from an invalid to a valid
+	// configuration or vice versa.
+	clf.Status.Synchronize(cls)
+	defer func() {
+		if err := r.Client.Status().Update(ctx, &clf); err != nil {
+			log.Error(err, "Error while updating status for ClusterLoggingForwarder %s/%s", clf.Namespace, clf.Name)
+		}
+	}()
 
 	clusterVersion, clusterID := r.GetClusterVersionID(ctx, request.Namespace)
 	resourceNames := factory.GenerateResourceNames(clf)

test/e2e/logforwarding/miscellaneous/forward_miscellaneous_test.go

@@ -3,8 +3,11 @@ package miscellaneous
 import (
 	"fmt"
 	"testing"
+	"time"
 
 	"github.com/openshift/cluster-logging-operator/test/helpers"
+	"k8s.io/apimachinery/pkg/util/wait"
+	"k8s.io/client-go/util/retry"
 
 	"github.com/stretchr/testify/require"
 	"golang.org/x/sync/errgroup"
@@ -43,9 +46,11 @@ var spec = loggingv1.ClusterLogForwarderSpec{
 	},
 }
 
-// TestLogForwardingWithEmptyCollection tests for issue https://github.com/openshift/cluster-logging-operator/issues/2312.
-// It creates a CL with cl.Spec.Collection set to nil. This would trigger a nil pointer exception without a
+// TestLogForwardingWithEmptyCollection tests for issues https://github.com/openshift/cluster-logging-operator/issues/2312
+// and https://github.com/openshift/cluster-logging-operator/issues/2314.
+// It first creates a CL with cl.Spec.Collection set to nil. This would trigger a nil pointer exception without a
 // fix in place.
+// It then updates the CL to a valid status. Without a fix in place, the CLF's status would not update.
 func TestLogForwardingWithEmptyCollection(t *testing.T) {
 	// First, make sure that the Operator can handle a nil cl.Spec.Collection.
 	// https://github.com/openshift/cluster-logging-operator/issues/2312
@@ -67,6 +72,31 @@ func TestLogForwardingWithEmptyCollection(t *testing.T) {
 	require.NoError(t, g.Wait())
 	require.NoError(t, c.WaitFor(clf, client.ClusterLogForwarderValidationFailure))
 	require.NoError(t, framework.WaitFor(helpers.ComponentTypeCollector))
+
+	// Now, make sure that the CLF's status updates to Ready when we update the CL resource to a valid status.
+	// https://github.com/openshift/cluster-logging-operator/issues/2314
+	t.Log("TestLogForwardingWithEmptyCollection: Make sure CLF updates when CL transitions to good state")
+	clSpec := &loggingv1.CollectionSpec{
+		Type:          loggingv1.LogCollectionTypeVector,
+		CollectorSpec: loggingv1.CollectorSpec{},
+	}
+	retryErr := retry.RetryOnConflict(retry.DefaultRetry, func() error {
+		if err := c.Get(cl); err != nil {
+			return err
+		}
+		cl.Spec.Collection = clSpec
+		return c.Update(cl)
+	})
+	require.NoError(t, retryErr)
+	// WaitFor alone will return too early and return an error. Instead, make use of the K8s retry framework and retry
+	// up to 30 seconds.
+	retryErr = retry.OnError(
+		wait.Backoff{Steps: 10, Duration: 3 * time.Second, Factor: 1.0},
+		func(error) bool { return true },
+		func() error { t.Log("Retrieving CLF status"); return c.WaitFor(clf, client.ClusterLogForwarderReady) },
+	)
+	require.NoError(t, retryErr)
+	require.NoError(t, framework.WaitFor(helpers.ComponentTypeCollector))
 }
 
 // TestLogForwardingWithEmptyCollection tests for issue https://github.com/openshift/cluster-logging-operator/issues/2315.