Clean up MedusaBackupJobs and propagate MedusaTask labels (#1339) (#1347

)

Co-authored-by: Clément Doucy <[email protected]>

CHANGELOG/CHANGELOG-1.17.md

@@ -15,6 +15,8 @@ When cutting a new release, update the `unreleased` heading to the tag being gen
 
 ## unreleased
 
+* [BUGFIX] [#1334](https://github.com/k8ssandra/k8ssandra-operator/issues/1334) Delete `MedusaBackupJobs` during purge
+* [BUGFIX] [#1336](https://github.com/k8ssandra/k8ssandra-operator/issues/1336) Propagate labels to the sync `MedusaTask` created after purge
 * [CHANGE] []() Update cass-operator to v1.21.0
 * [CHANGE] [#1313](https://github.com/k8ssandra/k8ssandra-operator/issues/1313) upgrade controller-runtime to 1.17 series, Go to 1.21.
 * [BUGFIX] [#1317](https://github.com/k8ssandra/k8ssandra-operator/issues/1317) Fix issues with caches in cluster scoped deployments where they were continuing to use a multi-namespace scoped cache and not an informer cache.

controllers/medusa/medusabackupjob_controller_test.go

@@ -34,6 +34,7 @@ const (
 	fakeBackupFileCount = int64(13)
 	fakeBackupByteSize  = int64(42)
 	fakeBackupHumanSize = "42.00 B"
+	fakeMaxBackupCount  = 1
 )
 
 func testMedusaBackupDatacenter(t *testing.T, ctx context.Context, f *framework.Framework, namespace string) {
@@ -392,8 +393,13 @@ func (c *fakeMedusaClient) BackupStatus(ctx context.Context, name string) (*medu
 }
 
 func (c *fakeMedusaClient) PurgeBackups(ctx context.Context) (*medusa.PurgeBackupsResponse, error) {
+	size := len(c.RequestedBackups)
+	if size > fakeMaxBackupCount {
+		c.RequestedBackups = c.RequestedBackups[size-fakeMaxBackupCount:]
+	}
+
 	response := &medusa.PurgeBackupsResponse{
-		NbBackupsPurged:           2,
+		NbBackupsPurged:           int32(size - len(c.RequestedBackups)),
 		NbObjectsPurged:           10,
 		TotalObjectsWithinGcGrace: 0,
 		TotalPurgedSize:           1000,

controllers/medusa/medusatask_controller.go

@@ -301,6 +301,20 @@ func (r *MedusaTaskReconciler) syncOperation(ctx context.Context, task *medusav1
 						return ctrl.Result{}, err
 					} else {
 						logger.Info("Deleted Cassandra Backup", "Backup", backup.ObjectMeta.Name)
+
+						backupJob := medusav1alpha1.MedusaBackupJob{
+							ObjectMeta: metav1.ObjectMeta{
+								Name:      backup.GetName(),
+								Namespace: backup.GetNamespace(),
+							},
+						}
+						logger.Info("Deleting MedusaBackupJob", "MedusaBackupJob", backupJob.GetName())
+
+						if err := r.Delete(ctx, &backupJob); err != nil && !errors.IsNotFound(err) {
+							logger.Error(err, "failed to delete MedusaBackupJob", "MedusaBackupJob", backupJob.GetName())
+						} else {
+							logger.Info("Deleted MedusaBackupJob", "MedusaBackupJob", backupJob.GetName())
+						}
 					}
 				}
 			}
@@ -377,6 +391,7 @@ func (r *MedusaTaskReconciler) scheduleSyncForPurge(task *medusav1alpha1.MedusaT
 				ObjectMeta: metav1.ObjectMeta{
 					Name:      task.GetObjectMeta().GetName() + "-sync",
 					Namespace: task.Namespace,
+					Labels:    task.GetLabels(),
 				},
 				Spec: medusav1alpha1.MedusaTaskSpec{
 					Operation:           medusav1alpha1.OperationTypeSync,

controllers/medusa/medusatask_controller_test.go

@@ -153,13 +153,19 @@ func testMedusaTasks(t *testing.T, ctx context.Context, f *framework.Framework,
 	backup4Created := createAndVerifyMedusaBackup(dc2Key, dc2, f, ctx, require, t, namespace, backup4)
 	require.True(backup4Created, "failed to create backup4")
 
+	// Ensure that 4 backups and backup jobs were created
+	checkBackupsAndJobs(require, ctx, 4, namespace, f, []string{})
+
 	// Purge backups and verify that only one out of three remains
 	t.Log("purge backups")
 
 	purgeTask := &api.MedusaTask{
 		ObjectMeta: metav1.ObjectMeta{
 			Namespace: namespace,
 			Name:      "purge-backups",
+			Labels: map[string]string{
+				"app": "medusa",
+			},
 		},
 		Spec: api.MedusaTaskSpec{
 			CassandraDatacenter: "dc1",
@@ -193,9 +199,18 @@ func testMedusaTasks(t *testing.T, ctx context.Context, f *framework.Framework,
 			return false
 		}
 
+		v, ok := updated.Labels["app"]
+		if !ok || v != "medusa" {
+			return false
+		}
+
 		return !updated.Status.FinishTime.IsZero()
 	}, timeout, interval)
 
+	// Ensure that 2 backups and backup jobs were deleted
+	deletedBackups := []string{backup1, backup2}
+	checkBackupsAndJobs(require, ctx, 2, namespace, f, deletedBackups)
+
 	medusaBackup4Key := framework.NewClusterKey(f.DataPlaneContexts[0], namespace, backup4)
 	medusaBackup4 := &api.MedusaBackup{}
 	err = f.Get(ctx, medusaBackup4Key, medusaBackup4)
@@ -206,3 +221,20 @@ func testMedusaTasks(t *testing.T, ctx context.Context, f *framework.Framework,
 	verifyObjectDoesNotExist(ctx, t, f, dc1Key, &cassdcapi.CassandraDatacenter{})
 	verifyObjectDoesNotExist(ctx, t, f, dc2Key, &cassdcapi.CassandraDatacenter{})
 }
+
+func checkBackupsAndJobs(require *require.Assertions, ctx context.Context, expectedLen int, namespace string, f *framework.Framework, deleted []string) {
+	var backups api.MedusaBackupList
+	err := f.List(ctx, framework.NewClusterKey(f.DataPlaneContexts[0], namespace, "list-backups"), &backups)
+	require.NoError(err, "failed to list medusabackup")
+	require.Len(backups.Items, expectedLen, "expected %d backups, got %d", expectedLen, len(backups.Items))
+
+	var jobs api.MedusaBackupJobList
+	err = f.List(ctx, framework.NewClusterKey(f.DataPlaneContexts[0], namespace, "list-backup-jobs"), &jobs)
+	require.NoError(err, "failed to list medusabackupjobs")
+	require.Len(jobs.Items, expectedLen, "expected %d jobs, got %d", expectedLen, len(jobs.Items))
+
+	for _, d := range deleted {
+		require.NotContains(backups.Items, d, "MedusaBackup %s to have been deleted", d)
+		require.NotContains(jobs.Items, d, "MedusaBackupJob %s to have been deleted", d)
+	}
+}