Improving Postgres CR Status with Additional Details

manifests/minimal-postgres-manifest.yaml

@@ -2,6 +2,8 @@ apiVersion: "acid.zalan.do/v1"
 kind: postgresql
 metadata:
   name: acid-minimal-cluster
+  labels:
+    cluster-name: acid-minimal-cluster
 spec:
   teamId: "acid"
   volume:

manifests/postgresql.crd.yaml

@@ -20,6 +20,10 @@ spec:
     storage: true
     subresources:
       status: {}
+      scale:
+          specReplicasPath: .spec.numberOfInstances
+          statusReplicasPath: .status.numberOfInstances
+          labelSelectorPath: .status.labelSelector
     additionalPrinterColumns:
     - name: Team
       type: string
@@ -51,7 +55,7 @@ spec:
     - name: Status
       type: string
       description: Current sync status of postgresql resource
-      jsonPath: .status.PostgresClusterStatus
+      jsonPath: .status.postgresClusterStatus
     schema:
       openAPIV3Schema:
         type: object
@@ -677,5 +681,30 @@ spec:
                     type: integer
           status:
             type: object
-            additionalProperties:
-              type: string
+            properties:
+              postgresClusterStatus:
+                type: string
+              numberOfInstances:
+                format: int32
+                type: integer
+              labelSelector:
+                type: string
+              observedGeneration:
+                format: int64
+                type: integer
+              conditions:
+                type: array
+                items:
+                  type: object
+                  properties:
+                    type:
+                      type: string
+                    status:
+                      type: string
+                    lastTransitionTime:
+                      type: string
+                      format: date-time
+                    reason:
+                      type: string
+                    message:
+                      type: string

pkg/apis/acid.zalan.do/v1/crds.go

@@ -25,6 +25,12 @@ const (
 	OperatorConfigCRDResourceShort  = "opconfig"
 )
 
+var (
+	specReplicasPath   = ".spec.numberOfInstances"
+	statusReplicasPath = ".status.numberOfInstances"
+	labelSelectorPath  = ".status.labelSelector"
+)
+
 // PostgresCRDResourceColumns definition of AdditionalPrinterColumns for postgresql CRD
 var PostgresCRDResourceColumns = []apiextv1.CustomResourceColumnDefinition{
 	{
@@ -72,7 +78,7 @@ var PostgresCRDResourceColumns = []apiextv1.CustomResourceColumnDefinition{
 		Name:        "Status",
 		Type:        "string",
 		Description: "Current sync status of postgresql resource",
-		JSONPath:    ".status.PostgresClusterStatus",
+		JSONPath:    ".status.postgresClusterStatus",
 	},
 }
 
@@ -1106,10 +1112,47 @@ var PostgresCRDResourceValidation = apiextv1.CustomResourceValidation{
 			},
 			"status": {
 				Type: "object",
-				AdditionalProperties: &apiextv1.JSONSchemaPropsOrBool{
-					Schema: &apiextv1.JSONSchemaProps{
+				Properties: map[string]apiextv1.JSONSchemaProps{
+					"postgresClusterStatus": {
 						Type: "string",
 					},
+					"numberOfInstances": {
+						Type:   "integer",
+						Format: "int32",
+					},
+					"labelSelector": {
+						Type: "string",
+					},
+					"observedGeneration": {
+						Type:   "integer",
+						Format: "int64",
+					},
+					"conditions": {
+						Type: "array",
+						Items: &apiextv1.JSONSchemaPropsOrArray{
+							Schema: &apiextv1.JSONSchemaProps{
+								Type: "object",
+								Properties: map[string]apiextv1.JSONSchemaProps{
+									"type": {
+										Type: "string",
+									},
+									"status": {
+										Type: "string",
+									},
+									"lastTransitionTime": {
+										Type:   "string",
+										Format: "date-time",
+									},
+									"reason": {
+										Type: "string",
+									},
+									"message": {
+										Type: "string",
+									},
+								},
+							},
+						},
+					},
 				},
 			},
 		},
@@ -1983,7 +2026,7 @@ var OperatorConfigCRDResourceValidation = apiextv1.CustomResourceValidation{
 func buildCRD(name, kind, plural, list, short string,
 	categories []string,
 	columns []apiextv1.CustomResourceColumnDefinition,
-	validation apiextv1.CustomResourceValidation) *apiextv1.CustomResourceDefinition {
+	validation apiextv1.CustomResourceValidation, specReplicasPath string, statusReplicasPath string, labelSelectorPath string) *apiextv1.CustomResourceDefinition {
 	return &apiextv1.CustomResourceDefinition{
 		TypeMeta: metav1.TypeMeta{
 			APIVersion: fmt.Sprintf("%s/%s", apiextv1.GroupName, apiextv1.SchemeGroupVersion.Version),
@@ -2010,6 +2053,11 @@ func buildCRD(name, kind, plural, list, short string,
 					Storage: true,
 					Subresources: &apiextv1.CustomResourceSubresources{
 						Status: &apiextv1.CustomResourceSubresourceStatus{},
+						Scale: &apiextv1.CustomResourceSubresourceScale{
+							SpecReplicasPath:   specReplicasPath,
+							StatusReplicasPath: statusReplicasPath,
+							LabelSelectorPath:  &labelSelectorPath,
+						},
 					},
 					AdditionalPrinterColumns: columns,
 					Schema:                   &validation,
@@ -2028,7 +2076,10 @@ func PostgresCRD(crdCategories []string) *apiextv1.CustomResourceDefinition {
 		PostgresCRDResourceShort,
 		crdCategories,
 		PostgresCRDResourceColumns,
-		PostgresCRDResourceValidation)
+		PostgresCRDResourceValidation,
+		specReplicasPath,
+		statusReplicasPath,
+		labelSelectorPath)
 }
 
 // ConfigurationCRD returns CustomResourceDefinition built from OperatorConfigCRDResource
@@ -2040,5 +2091,8 @@ func ConfigurationCRD(crdCategories []string) *apiextv1.CustomResourceDefinition
 		OperatorConfigCRDResourceShort,
 		crdCategories,
 		OperatorConfigCRDResourceColumns,
-		OperatorConfigCRDResourceValidation)
+		OperatorConfigCRDResourceValidation,
+		specReplicasPath,
+		statusReplicasPath,
+		labelSelectorPath)
 }

pkg/apis/acid.zalan.do/v1/postgresql_type.go

@@ -3,6 +3,7 @@ package v1
 // Postgres CRD definition, please use CamelCase for field names.
 
 import (
+	"k8s.io/apimachinery/pkg/api/equality"
 	"time"
 
 	v1 "k8s.io/api/core/v1"
@@ -225,9 +226,48 @@ type Sidecar struct {
 // UserFlags defines flags (such as superuser, nologin) that could be assigned to individual users
 type UserFlags []string
 
+type Conditions []Condition
+
+type ConditionType string
+type VolatileTime struct {
+	Inner metav1.Time `json:",inline"`
+}
+
+// MarshalJSON implements the json.Marshaler interface.
+func (t VolatileTime) MarshalJSON() ([]byte, error) {
+	return t.Inner.MarshalJSON()
+}
+
+// UnmarshalJSON implements the json.Unmarshaller interface.
+func (t *VolatileTime) UnmarshalJSON(b []byte) error {
+	return t.Inner.UnmarshalJSON(b)
+}
+
+func init() {
+	equality.Semantic.AddFunc(
+		// Always treat VolatileTime fields as equivalent.
+		func(VolatileTime, VolatileTime) bool {
+			return true
+		},
+	)
+}
+
+// Condition contains the conditions of the PostgreSQL cluster
+type Condition struct {
+	Type               ConditionType      `json:"type" description:"type of status condition"`
+	Status             v1.ConditionStatus `json:"status" description:"status of the condition, one of True, False, Unknown"`
+	LastTransitionTime VolatileTime       `json:"lastTransitionTime,omitempty" description:"last time the condition transit from one status to another"`
+	Reason             string             `json:"reason,omitempty" description:"one-word CamelCase reason for the condition's last transition"`
+	Message            string             `json:"message,omitempty" description:"human-readable message indicating details about last transition"`
+}
+
 // PostgresStatus contains status of the PostgreSQL cluster (running, creation failed etc.)
 type PostgresStatus struct {
-	PostgresClusterStatus string `json:"PostgresClusterStatus"`
+	PostgresClusterStatus string     `json:"postgresClusterStatus"`
+	NumberOfInstances     int32      `json:"numberOfInstances"`
+	LabelSelector         string     `json:"labelSelector"`
+	ObservedGeneration    int64      `json:"observedGeneration,omitempty"`
+	Conditions            Conditions `json:"conditions,omitempty"`
 }
 
 // ConnectionPooler Options for connection pooler

pkg/apis/acid.zalan.do/v1/util.go

@@ -100,7 +100,3 @@ func (postgresStatus PostgresStatus) Running() bool {
 func (postgresStatus PostgresStatus) Creating() bool {
 	return postgresStatus.PostgresClusterStatus == ClusterStatusCreating
 }
-
-func (postgresStatus PostgresStatus) String() string {
-	return postgresStatus.PostgresClusterStatus
-}

pkg/cluster/cluster.go

@@ -255,16 +255,20 @@ func (c *Cluster) Create() (err error) {
 		ss             *appsv1.StatefulSet
 	)
 
+	//Even though its possible to propogate other CR labels to the pods, picking the default label here since its propogated to all the pods by default. But this means that in order for the scale subresource to work properly, user must set the "cluster-name" key in their CRs with value matching the CR name.
+	labelstring := fmt.Sprintf("%s=%s", "cluster-name", c.Postgresql.ObjectMeta.Labels["cluster-name"]) //TODO: make this configurable.
+
 	defer func() {
 		var (
 			pgUpdatedStatus *acidv1.Postgresql
 			errStatus       error
 		)
+		existingConditions := c.Postgresql.Status.Conditions
 		if err == nil {
-			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning) //TODO: are you sure it's running?
+			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning, c.Postgresql.Spec.NumberOfInstances, labelstring, c.Postgresql.Generation, existingConditions, "") //TODO: are you sure it's running?
 		} else {
 			c.logger.Warningf("cluster created failed: %v", err)
-			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusAddFailed)
+			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusAddFailed, 0, labelstring, 0, existingConditions, err.Error())
 		}
 		if errStatus != nil {
 			c.logger.Warningf("could not set cluster status: %v", errStatus)
@@ -274,7 +278,8 @@ func (c *Cluster) Create() (err error) {
 		}
 	}()
 
-	pgCreateStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusCreating)
+	existingConditions := c.Postgresql.Status.Conditions
+	pgCreateStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusCreating, 0, labelstring, 0, existingConditions, "")
 	if err != nil {
 		return fmt.Errorf("could not set cluster status: %v", err)
 	}
@@ -927,7 +932,9 @@ func (c *Cluster) Update(oldSpec, newSpec *acidv1.Postgresql) error {
 	c.mu.Lock()
 	defer c.mu.Unlock()
 
-	c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusUpdating)
+	labelstring := fmt.Sprintf("%s=%s", "cluster-name", c.Postgresql.ObjectMeta.Labels["cluster-name"])
+	existingConditions := c.Postgresql.Status.Conditions
+	c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusUpdating, c.Postgresql.Status.NumberOfInstances, labelstring, c.Postgresql.Status.ObservedGeneration, existingConditions, "")
 	c.setSpec(newSpec)
 
 	defer func() {
@@ -936,9 +943,9 @@ func (c *Cluster) Update(oldSpec, newSpec *acidv1.Postgresql) error {
 			err             error
 		)
 		if updateFailed {
-			pgUpdatedStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusUpdateFailed)
+			pgUpdatedStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusUpdateFailed, c.Postgresql.Status.NumberOfInstances, labelstring, c.Postgresql.Status.ObservedGeneration, existingConditions, err.Error())
 		} else {
-			pgUpdatedStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning)
+			pgUpdatedStatus, err = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning, newSpec.Spec.NumberOfInstances, labelstring, c.Postgresql.Generation, existingConditions, "")
 		}
 		if err != nil {
 			c.logger.Warningf("could not set cluster status: %v", err)

pkg/cluster/sync.go

@@ -46,11 +46,13 @@ func (c *Cluster) Sync(newSpec *acidv1.Postgresql) error {
 			pgUpdatedStatus *acidv1.Postgresql
 			errStatus       error
 		)
+		labelstring := fmt.Sprintf("%s=%s", "cluster-name", c.Postgresql.ObjectMeta.Labels["cluster-name"])
+		existingConditions := c.Postgresql.Status.Conditions
 		if err != nil {
 			c.logger.Warningf("error while syncing cluster state: %v", err)
-			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusSyncFailed)
+			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusSyncFailed, newSpec.Status.NumberOfInstances, labelstring, c.Postgresql.Status.ObservedGeneration, existingConditions, errStatus.Error())
 		} else if !c.Status.Running() {
-			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning)
+			pgUpdatedStatus, errStatus = c.KubeClient.SetPostgresCRDStatus(c.clusterName(), acidv1.ClusterStatusRunning, newSpec.Spec.NumberOfInstances, labelstring, c.Postgresql.Generation, existingConditions, "")
 		}
 		if errStatus != nil {
 			c.logger.Warningf("could not set cluster status: %v", errStatus)

pkg/controller/postgresql.go

@@ -161,7 +161,9 @@
 func (c *Controller) addCluster(lg *logrus.Entry, clusterName spec.NamespacedName, pgSpec *acidv1.Postgresql) (*cluster.Cluster, error) {
 	if c.opConfig.EnableTeamIdClusternamePrefix {
 		if _, err := acidv1.ExtractClusterName(clusterName.Name, pgSpec.Spec.TeamID); err != nil {
-			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusInvalid)
+			labelstring := fmt.Sprintf("%s=%s", "cluster-name", pgSpec.ObjectMeta.Labels["cluster-name"])
+			existingConditions := pgSpec.Status.Conditions
+			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusInvalid, pgSpec.Status.NumberOfInstances, labelstring, pgSpec.Status.ObservedGeneration, existingConditions, err.Error())
 			return nil, err
 		}
 	}
@@ -207,10 +209,10 @@
 	if event.EventType == EventRepair {
 		runRepair, lastOperationStatus := cl.NeedsRepair()
 		if !runRepair {
 			lg.Debugf("observed cluster status %s, repair is not required", lastOperationStatus)
 			return
 		}
 		lg.Debugf("observed cluster status %s, running sync scan to repair the cluster", lastOperationStatus)
 		event.EventType = EventSync
 	}
 
@@ -472,16 +474,18 @@
 
 	if clusterError != "" && eventType != EventDelete {
 		c.logger.WithField("cluster-name", clusterName).Debugf("skipping %q event for the invalid cluster: %s", eventType, clusterError)
+		labelstring := fmt.Sprintf("%s=%s", "cluster-name", informerNewSpec.ObjectMeta.Labels["cluster-name"])
+		existingConditions := informerNewSpec.Status.Conditions
 
 		switch eventType {
 		case EventAdd:
-			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusAddFailed)
+			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusAddFailed, informerNewSpec.Status.NumberOfInstances, labelstring, informerNewSpec.Status.ObservedGeneration, existingConditions, clusterError)
 			c.eventRecorder.Eventf(c.GetReference(informerNewSpec), v1.EventTypeWarning, "Create", "%v", clusterError)
 		case EventUpdate:
-			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusUpdateFailed)
+			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusUpdateFailed, informerNewSpec.Status.NumberOfInstances, labelstring, informerNewSpec.Status.ObservedGeneration, existingConditions, clusterError)
 			c.eventRecorder.Eventf(c.GetReference(informerNewSpec), v1.EventTypeWarning, "Update", "%v", clusterError)
 		default:
-			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusSyncFailed)
+			c.KubeClient.SetPostgresCRDStatus(clusterName, acidv1.ClusterStatusSyncFailed, informerNewSpec.Status.NumberOfInstances, labelstring, informerNewSpec.Status.ObservedGeneration, existingConditions, clusterError)
 			c.eventRecorder.Eventf(c.GetReference(informerNewSpec), v1.EventTypeWarning, "Sync", "%v", clusterError)
 		}