cass-operator v1.23.0 (#1461)

* cass-operator v1.23.0 API changes, such as SanitiziedName -> LabelResourceName

Add updated CRD and exposed method

Change name generation to follow the real naming from cass-operator

Update tag to 1.23.0 and fix medusa tests

Change DcPrefix() in the test

Some resourceName prefix check fixes

Change schemas.go check to use CassandraDatacenter

Fix status updates and add a new status field, contextName for the Datacenter

Change back slower polling intervals for the tests, or they start to fail.

* Update CHANGELOG

CHANGELOG/CHANGELOG-1.21.md

@@ -15,8 +15,11 @@ When cutting a new release, update the `unreleased` heading to the tag being gen
 
 ## unreleased
 
+* [CHANGE] [#1450](https://github.com/k8ssandra/k8ssandra-operator/issues/1450) Update datacenter labels to use Kubernetes resource names for CassandraDatacenter, not the cleaned override name. Update to cass-operator 1.23.0
 * [CHANGE] [#1441](https://github.com/k8ssandra/k8ssandra-operator/issues/1441) Use k8ssandra-client instead of k8ssandra-tools for CRD upgrades
-* [BUGFIX] [#1383](https://github.com/k8ssandra/k8ssandra-operator/issues/1383) Do not create MedusaBackup if MedusaBakupJob did not fully succeed
 * [ENHANCEMENT] [#1667](https://github.com/k8ssahttps://github.com/k8ssandra/k8ssandra/issues/1667) Add `skipSchemaMigration` option to `K8ssandraCluster.spec.reaper`
 * [FEATURE] [#1034](https://github.com/k8ssandra/k8ssandra-operator/issues/1034) Add support for priorityClassName
 * [BUGFIX] [#1454](https://github.com/k8ssandra/k8ssandra-operator/issues/1454) Do not try to work out backup status if there are no pods
+* [BUGFIX] [#1383](https://github.com/k8ssandra/k8ssandra-operator/issues/1383) Do not create MedusaBackup if MedusaBakupJob did not fully succeed
+* [BUGFIX] [#1460](https://github.com/k8ssandra/k8ssandra-operator/issues/1460) Fix podName calculations in medusa's hostmap.go to account for unbalanced racks also
+* [BUGFIX] [#1466](https://github.com/k8ssandra/k8ssandra-operator/issues/1466) Do not overwrite existing status fields or forget to write the changes. Also, add new ContextName for the Datacenter to know where it used to be. 

apis/k8ssandra/v1alpha1/k8ssandracluster_types.go

@@ -143,6 +143,7 @@ type K8ssandraClusterCondition struct {
 
 // K8ssandraStatus defines the observed of a k8ssandra instance
 type K8ssandraStatus struct {
+	ContextName          string                               `json:"contextName,omitempty"`
 	DecommissionProgress DecommissionProgress                 `json:"decommissionProgress,omitempty"`
 	Cassandra            *cassdcapi.CassandraDatacenterStatus `json:"cassandra,omitempty"`
 	Stargate             *stargateapi.StargateStatus          `json:"stargate,omitempty"`
@@ -443,8 +444,7 @@ type DatacenterOptions struct {
 	// The k8s service account to use for the Cassandra pods
 	ServiceAccount string `json:"serviceAccount,omitempty"`
 
-	// DatacenterName allows to override the name of the Cassandra datacenter. Kubernetes objects will be named after a sanitized version of it if set, and if not metadata.name. In Cassandra the DC name will be overridden by this value.
-	// It may generate some confusion as objects created for the DC will have a different name than the CasandraDatacenter object itself.
+	// DatacenterName allows to override the name of the Cassandra datacenter. In Cassandra the DC name will be overridden by this value.
 	// This setting can create conflicts if multiple DCs coexist in the same namespace if metadata.name for a DC with no override is set to the same value as the override name of another DC.
 	// Use cautiously.
 	// +optional

apis/k8ssandra/v1alpha1/k8ssandracluster_webhook.go

@@ -21,6 +21,8 @@ import (
 	"strings"
 
 	"github.com/Masterminds/semver/v3"
+	cassdcapi "github.com/k8ssandra/cass-operator/apis/cassandra/v1beta1"
+	"github.com/k8ssandra/cass-operator/pkg/reconciliation"
 	reaperapi "github.com/k8ssandra/k8ssandra-operator/apis/reaper/v1alpha1"
 
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -132,15 +134,31 @@ func (r *K8ssandraCluster) validateK8ssandraCluster() error {
 }
 
 func (r *K8ssandraCluster) validateStatefulsetNameSize() error {
+	clusterName := r.ObjectMeta.Name
+	if r.Spec.Cassandra.ClusterName != "" {
+		clusterName = r.Spec.Cassandra.ClusterName
+	}
+
 	for _, dc := range r.Spec.Cassandra.Datacenters {
+		realDc := &cassdcapi.CassandraDatacenter{
+			ObjectMeta: metav1.ObjectMeta{
+				Name: dc.Meta.Name,
+			},
+			Spec: cassdcapi.CassandraDatacenterSpec{
+				ClusterName: clusterName,
+			},
+		}
+
 		if len(dc.Racks) > 0 {
 			for _, rack := range dc.Racks {
-				if len(r.SanitizedName()+"-"+dc.CassDcName()+"-"+rack.Name+"-sts-") > 60 {
+				stsName := reconciliation.NewNamespacedNameForStatefulSet(realDc, rack.Name)
+				if len(stsName.Name) > 60 {
 					return fmt.Errorf("the name of the statefulset for rack %s in DC %s is too long", rack.Name, dc.CassDcName())
 				}
 			}
 		} else {
-			if len(r.SanitizedName()+"-"+dc.CassDcName()+"-default-sts-") > 60 {
+			stsName := reconciliation.NewNamespacedNameForStatefulSet(realDc, "default")
+			if len(stsName.Name) > 60 {
 				return fmt.Errorf("the name of the statefulset for DC %s is too long", dc.CassDcName())
 			}
 		}

charts/k8ssandra-operator/crds/k8ssandra-operator-crds.yaml

@@ -2222,8 +2222,7 @@ spec:
                     type: array
                   datacenterName:
                     description: |-
-                      DatacenterName allows to override the name of the Cassandra datacenter. Kubernetes objects will be named after a sanitized version of it if set, and if not metadata.name. In Cassandra the DC name will be overridden by this value.
-                      It may generate some confusion as objects created for the DC will have a different name than the CasandraDatacenter object itself.
+                      DatacenterName allows to override the name of the Cassandra datacenter. In Cassandra the DC name will be overridden by this value.
                       This setting can create conflicts if multiple DCs coexist in the same namespace if metadata.name for a DC with no override is set to the same value as the override name of another DC.
                       Use cautiously.
                     type: string
@@ -4251,8 +4250,7 @@ spec:
                           type: array
                         datacenterName:
                           description: |-
-                            DatacenterName allows to override the name of the Cassandra datacenter. Kubernetes objects will be named after a sanitized version of it if set, and if not metadata.name. In Cassandra the DC name will be overridden by this value.
-                            It may generate some confusion as objects created for the DC will have a different name than the CasandraDatacenter object itself.
+                            DatacenterName allows to override the name of the Cassandra datacenter. In Cassandra the DC name will be overridden by this value.
                             This setting can create conflicts if multiple DCs coexist in the same namespace if metadata.name for a DC with no override is set to the same value as the override name of another DC.
                             Use cautiously.
                           type: string
@@ -31083,6 +31081,9 @@ spec:
                             with the management API
                           format: date-time
                           type: string
+                        metadataVersion:
+                          format: int64
+                          type: integer
                         nodeReplacements:
                           items:
                             type: string
@@ -31183,6 +31184,8 @@ spec:
                           format: date-time
                           type: string
                       type: object
+                    contextName:
+                      type: string
                     decommissionProgress:
                       type: string
                     reaper:

config/cass-operator/cluster-scoped/kustomization.yaml

@@ -2,4 +2,4 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 
 resources:
-- github.com/k8ssandra/cass-operator/config/deployments/cluster?ref=v1.22.4
+- github.com/k8ssandra/cass-operator/config/deployments/cluster?ref=v1.23.0

config/cass-operator/ns-scoped/kustomization.yaml

@@ -2,4 +2,4 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 
 resources:
-- github.com/k8ssandra/cass-operator/config/deployments/default?ref=v1.22.4
+- github.com/k8ssandra/cass-operator/config/deployments/default?ref=v1.23.0

config/crd/bases/k8ssandra.io_k8ssandraclusters.yaml

@@ -2160,8 +2160,7 @@ spec:
                     type: array
                   datacenterName:
                     description: |-
-                      DatacenterName allows to override the name of the Cassandra datacenter. Kubernetes objects will be named after a sanitized version of it if set, and if not metadata.name. In Cassandra the DC name will be overridden by this value.
-                      It may generate some confusion as objects created for the DC will have a different name than the CasandraDatacenter object itself.
+                      DatacenterName allows to override the name of the Cassandra datacenter. In Cassandra the DC name will be overridden by this value.
                       This setting can create conflicts if multiple DCs coexist in the same namespace if metadata.name for a DC with no override is set to the same value as the override name of another DC.
                       Use cautiously.
                     type: string
@@ -4189,8 +4188,7 @@ spec:
                           type: array
                         datacenterName:
                           description: |-
-                            DatacenterName allows to override the name of the Cassandra datacenter. Kubernetes objects will be named after a sanitized version of it if set, and if not metadata.name. In Cassandra the DC name will be overridden by this value.
-                            It may generate some confusion as objects created for the DC will have a different name than the CasandraDatacenter object itself.
+                            DatacenterName allows to override the name of the Cassandra datacenter. In Cassandra the DC name will be overridden by this value.
                             This setting can create conflicts if multiple DCs coexist in the same namespace if metadata.name for a DC with no override is set to the same value as the override name of another DC.
                             Use cautiously.
                           type: string
@@ -31021,6 +31019,9 @@ spec:
                             with the management API
                           format: date-time
                           type: string
+                        metadataVersion:
+                          format: int64
+                          type: integer
                         nodeReplacements:
                           items:
                             type: string
@@ -31121,6 +31122,8 @@ spec:
                           format: date-time
                           type: string
                       type: object
+                    contextName:
+                      type: string
                     decommissionProgress:
                       type: string
                     reaper:

controllers/k8ssandra/cassandra_telemetry_reconciler.go

@@ -37,9 +37,9 @@ func (r *K8ssandraClusterReconciler) reconcileCassandraDCTelemetry(
 	cfg := telemetry.PrometheusResourcer{
 		MonitoringTargetNS:   actualDc.Namespace,
 		MonitoringTargetName: actualDc.Name,
-		ServiceMonitorName:   cassdcapi.CleanupForKubernetes(kc.CassClusterName() + "-" + actualDc.DatacenterName() + "-" + "cass-servicemonitor"),
+		ServiceMonitorName:   cassdcapi.CleanupForKubernetes(kc.CassClusterName() + "-" + actualDc.LabelResourceName() + "-" + "cass-servicemonitor"),
 		Logger:               logger,
-		CommonLabels:         mustLabels(kc.Name, kc.Namespace, actualDc.DatacenterName(), commonLabels),
+		CommonLabels:         mustLabels(kc.Name, kc.Namespace, actualDc.LabelResourceName(), commonLabels),
 	}
 	logger.Info("merged TelemetrySpec constructed", "mergedSpec", mergedSpec, "cluster", kc.Name)
 	// Confirm telemetry config is valid (e.g. Prometheus is installed if it is requested.)

controllers/k8ssandra/cleanup.go

@@ -144,7 +144,7 @@ func (r *K8ssandraClusterReconciler) checkFinalizer(ctx context.Context, kc *api
 }
 
 func (r *K8ssandraClusterReconciler) checkDcDeletion(ctx context.Context, kc *api.K8ssandraCluster, logger logr.Logger) result.ReconcileResult {
-	dcName, dcNameOverride := k8ssandra.GetDatacenterForDecommission(kc)
+	dcName := k8ssandra.GetDatacenterForDecommission(kc)
 	if dcName == "" {
 		return result.Continue()
 	}
@@ -163,76 +163,78 @@ func (r *K8ssandraClusterReconciler) checkDcDeletion(ctx context.Context, kc *ap
 	default:
 		logger.Info("Proceeding with DC deletion", "DC", dcName)
 
-		cassDcName := dcName
-		if dcNameOverride != "" {
-			cassDcName = dcNameOverride
-		}
-		return r.deleteDc(ctx, kc, dcName, cassDcName, logger)
+		return r.deleteDc(ctx, kc, dcName, logger)
 	}
 }
 
-func (r *K8ssandraClusterReconciler) deleteDc(ctx context.Context, kc *api.K8ssandraCluster, dcName string, cassDcName string, logger logr.Logger) result.ReconcileResult {
+func (r *K8ssandraClusterReconciler) deleteDc(ctx context.Context, kc *api.K8ssandraCluster, dcName string, logger logr.Logger) result.ReconcileResult {
 	kcKey := utils.GetKey(kc)
 
-	stargate, remoteClient, err := r.findStargateForDeletion(ctx, kcKey, cassDcName, nil)
+	dcRemoteClient, err := r.ClientCache.GetRemoteClient(kc.Status.Datacenters[dcName].ContextName)
+	if err != nil {
+		return result.Error(err)
+	}
+
+	dc, _, err := r.findDcForDeletion(ctx, kcKey, dcName, dcRemoteClient)
+	if err != nil {
+		return result.Error(err)
+	}
+
+	if dc == nil {
+		// Deletion was already done
+		delete(kc.Status.Datacenters, dcName)
+		logger.Info("DC deletion finished", "DC", dcName)
+		return result.Continue()
+	}
+
+	stargate, remoteClient, err := r.findStargateForDeletion(ctx, kcKey, dc.DatacenterName(), nil)
 	if err != nil {
 		return result.Error(err)
 	}
 
 	if stargate != nil {
 		if err = remoteClient.Delete(ctx, stargate); err != nil && !errors.IsNotFound(err) {
-			return result.Error(fmt.Errorf("failed to delete Stargate for dc (%s): %v", cassDcName, err))
+			return result.Error(fmt.Errorf("failed to delete Stargate for dc (%s): %v", dc.DatacenterName(), err))
 		}
 		logger.Info("Deleted Stargate", "Stargate", utils.GetKey(stargate))
 	}
 
-	reaper, remoteClient, err := r.findReaperForDeletion(ctx, kcKey, cassDcName, remoteClient)
+	reaper, remoteClient, err := r.findReaperForDeletion(ctx, kcKey, dc.DatacenterName(), remoteClient)
 	if err != nil {
 		return result.Error(err)
 	}
 
 	if reaper != nil {
 		if err = remoteClient.Delete(ctx, reaper); err != nil && !errors.IsNotFound(err) {
-			return result.Error(fmt.Errorf("failed to delete Reaper for dc (%s): %v", cassDcName, err))
+			return result.Error(fmt.Errorf("failed to delete Reaper for dc (%s): %v", dc.DatacenterName(), err))
 		}
 		logger.Info("Deleted Reaper", "Reaper", utils.GetKey(reaper))
 	}
 
-	dc, remoteClient, err := r.findDcForDeletion(ctx, kcKey, dcName, remoteClient)
-	if err != nil {
+	if err := r.deleteContactPointsService(ctx, kc, dc, logger); err != nil {
 		return result.Error(err)
 	}
 
-	if dc != nil {
-		if err := r.deleteContactPointsService(ctx, kc, dc, logger); err != nil {
-			return result.Error(err)
-		}
-
-		if dc.GetConditionStatus(cassdcapi.DatacenterDecommission) == corev1.ConditionTrue {
-			logger.Info("CassandraDatacenter decommissioning in progress", "CassandraDatacenter", utils.GetKey(dc))
-			// There is no need to requeue here. Reconciliation will be trigger by updates made by cass-operator.
-			return result.Done()
-		}
-
-		if !annotations.HasAnnotationWithValue(dc, cassdcapi.DecommissionOnDeleteAnnotation, "true") {
-			patch := client.MergeFrom(dc.DeepCopy())
-			annotations.AddAnnotation(dc, cassdcapi.DecommissionOnDeleteAnnotation, "true")
-			if err = remoteClient.Patch(ctx, dc, patch); err != nil {
-				return result.Error(fmt.Errorf("failed to add %s annotation to dc: %v", cassdcapi.DecommissionOnDeleteAnnotation, err))
-			}
-		}
-
-		if err = remoteClient.Delete(ctx, dc); err != nil && !errors.IsNotFound(err) {
-			return result.Error(fmt.Errorf("failed to delete CassandraDatacenter (%s): %v", dcName, err))
-		}
-		logger.Info("Deleted CassandraDatacenter", "CassandraDatacenter", utils.GetKey(dc))
+	if dc.GetConditionStatus(cassdcapi.DatacenterDecommission) == corev1.ConditionTrue {
+		logger.Info("CassandraDatacenter decommissioning in progress", "CassandraDatacenter", utils.GetKey(dc))
 		// There is no need to requeue here. Reconciliation will be trigger by updates made by cass-operator.
 		return result.Done()
 	}
 
-	delete(kc.Status.Datacenters, dcName)
-	logger.Info("DC deletion finished", "DC", dcName)
-	return result.Continue()
+	if !annotations.HasAnnotationWithValue(dc, cassdcapi.DecommissionOnDeleteAnnotation, "true") {
+		patch := client.MergeFrom(dc.DeepCopy())
+		annotations.AddAnnotation(dc, cassdcapi.DecommissionOnDeleteAnnotation, "true")
+		if err = dcRemoteClient.Patch(ctx, dc, patch); err != nil {
+			return result.Error(fmt.Errorf("failed to add %s annotation to dc: %v", cassdcapi.DecommissionOnDeleteAnnotation, err))
+		}
+	}
+
+	if err = dcRemoteClient.Delete(ctx, dc); err != nil && !errors.IsNotFound(err) {
+		return result.Error(fmt.Errorf("failed to delete CassandraDatacenter (%s): %v", dcName, err))
+	}
+	logger.Info("Deleted CassandraDatacenter", "CassandraDatacenter", utils.GetKey(dc))
+	// There is no need to requeue here. Reconciliation will be trigger by updates made by cass-operator.
+	return result.Done()
 }
 
 func (r *K8ssandraClusterReconciler) findStargateForDeletion(

controllers/k8ssandra/contact_points_service.go

@@ -2,6 +2,7 @@ package k8ssandra
 
 import (
 	"context"
+
 	"github.com/go-logr/logr"
 	cassdcapi "github.com/k8ssandra/cass-operator/apis/cassandra/v1beta1"
 	api "github.com/k8ssandra/k8ssandra-operator/apis/k8ssandra/v1alpha1"
@@ -82,7 +83,7 @@ func (r *K8ssandraClusterReconciler) loadAllPodsEndpoints(
 func contactPointsServiceKey(kc *api.K8ssandraCluster, dc *cassdcapi.CassandraDatacenter) client.ObjectKey {
 	return types.NamespacedName{
 		Namespace: kc.Namespace,
-		Name:      kc.SanitizedName() + "-" + dc.SanitizedName() + "-contact-points-service",
+		Name:      kc.SanitizedName() + "-" + dc.LabelResourceName() + "-contact-points-service",
 	}
 }
 

controllers/k8ssandra/datacenters.go

@@ -148,7 +148,7 @@ func (r *K8ssandraClusterReconciler) reconcileDatacenters(ctx context.Context, k
 				return result.Error(fmt.Errorf("CassandraDatacenter %s has cluster name %s, but expected %s. Cluster name cannot be changed in an existing cluster", dcKey, actualDc.Spec.ClusterName, cassClusterName)), actualDcs
 			}
 
-			r.setStatusForDatacenter(kc, actualDc)
+			r.setStatusForDatacenter(kc, actualDc, dcConfig.K8sContext)
 
 			r.reconcileContactPointsService(ctx, kc, actualDc, remoteClient, dcLogger)
 
@@ -309,7 +309,7 @@ func (r *K8ssandraClusterReconciler) reconcileDatacenters(ctx context.Context, k
 	return result.Continue(), actualDcs
 }
 
-func (r *K8ssandraClusterReconciler) setStatusForDatacenter(kc *api.K8ssandraCluster, dc *cassdcapi.CassandraDatacenter) {
+func (r *K8ssandraClusterReconciler) setStatusForDatacenter(kc *api.K8ssandraCluster, dc *cassdcapi.CassandraDatacenter, targetContext string) {
 	if len(kc.Status.Datacenters) == 0 {
 		kc.Status.Datacenters = make(map[string]api.K8ssandraStatus, 0)
 	}
@@ -318,9 +318,15 @@ func (r *K8ssandraClusterReconciler) setStatusForDatacenter(kc *api.K8ssandraClu
 
 	if found {
 		dc.Status.DeepCopyInto(kdcStatus.Cassandra)
+		if kdcStatus.ContextName != targetContext {
+			// This is pretty fatal situation if it happens to actually change the context, but for updates from previous versions we need it
+			kdcStatus.ContextName = targetContext
+		}
+		kc.Status.Datacenters[dc.Name] = kdcStatus
 	} else {
 		kc.Status.Datacenters[dc.Name] = api.K8ssandraStatus{
-			Cassandra: dc.Status.DeepCopy(),
+			ContextName: targetContext,
+			Cassandra:   dc.Status.DeepCopy(),
 		}
 	}
 }

controllers/k8ssandra/reaper.go

@@ -19,6 +19,7 @@ package k8ssandra
 import (
 	"context"
 	"fmt"
+
 	"github.com/go-logr/logr"
 	cassdcapi "github.com/k8ssandra/cass-operator/apis/cassandra/v1beta1"
 	api "github.com/k8ssandra/k8ssandra-operator/apis/k8ssandra/v1alpha1"
@@ -242,11 +243,8 @@ func (r *K8ssandraClusterReconciler) setStatusForReaper(kc *api.K8ssandraCluster
 
 func (r *K8ssandraClusterReconciler) removeReaperStatus(kc *api.K8ssandraCluster, dcName string) {
 	if kdcStatus, found := kc.Status.Datacenters[dcName]; found {
-		kc.Status.Datacenters[dcName] = api.K8ssandraStatus{
-			Reaper:    nil,
-			Cassandra: kdcStatus.Cassandra.DeepCopy(),
-			Stargate:  kdcStatus.Stargate.DeepCopy(),
-		}
+		kdcStatus.Reaper = nil
+		kc.Status.Datacenters[dcName] = kdcStatus
 	}
 }