SCF-830: Added state management for cluster hibernation

docs/reference/cluster_manifest.md

@@ -486,6 +486,25 @@ Note that `s3_wal_path` and `gs_wal_path` are mutually exclusive.
   from a remote primary. See the Patroni documentation
   [here](https://patroni.readthedocs.io/en/latest/standby_cluster.html) for more details. Optional.
 
+## Lifecycle configuration
+
+Parameters to control cluster hibernate/wake-up behavior.
+
+* **phase**
+  Set to `"stopped"` to hibernate the cluster. When this field is set on a
+  running cluster, the operator will:
+  * Store the current number of instances in the status
+  * Scale down the StatefulSet to 0 replicas
+  * Scale down the connection pooler to 0 replicas
+  * Set the cluster status to "Stopping", then "Stopped"
+
+  When this field is removed from a stopped cluster, the operator will:
+  * Restore the number of instances from the stored value
+  * Scale up the StatefulSet and connection pooler
+  * Set the cluster status to "Updating", then "Running"
+
+  This field is optional. When not set, the cluster operates normally.
+
 ## Volume properties
 
 Those parameters are grouped under the `volume` top-level key and define the
@@ -714,3 +733,21 @@ can have the following properties:
 
 * **memory**
   memory requests to be set as an annotation on the stream resource. Optional.
+
+## Status fields
+
+The operator reports the cluster state through the `status` sub-resource. These
+fields are managed by the operator and should not be set manually.
+
+* **PostgresClusterStatus**
+  Current state of the cluster. One of: Creating, Updating, Running,
+  UpdateFailed, SyncFailed, CreateFailed, Invalid, Stopping, Stopped.
+
+* **previousNumberOfInstances**
+  The number of instances the cluster had before hibernation. Used to restore
+  the cluster to its previous size when waking up. Cleared after wake-up.
+
+* **previousPoolerInstances**
+  A map of connection pooler role to its replica count before hibernation.
+  The keys are "master" and "replica". Used to restore the pooler when waking
+  up. Cleared after wake-up.

docs/user.md

@@ -930,6 +930,95 @@ When you apply this manifest, the operator will:
 
 The process is asynchronous. You can monitor the operator logs and the state of the `postgresql` resource to follow the progress. Once the new cluster is up and running, your applications can reconnect.
 
+## Hibernate and Wake-up a Cluster
+
+The operator supports hibernating a PostgreSQL cluster to save resources when it's
+not needed, and waking it up again when required. This feature:
+
+* Scales down the PostgreSQL StatefulSet to 0 replicas (stops all pods)
+* Scales down the connection pooler to 0 replicas
+* Preserves the cluster configuration and data (PVCs are retained)
+* Stores the previous replica counts for automatic restoration
+
+### Initiating Hibernate
+
+To hibernate a running cluster, set the `lifecycle.phase` field to `"stopped"`:
+
+```yaml
+apiVersion: "acid.zalan.do/v1"
+kind: postgresql
+metadata:
+  name: acid-test-cluster
+spec:
+  teamId: "test-team"
+  # ... other cluster parameters
+  numberOfInstances: 3
+  lifecycle:
+    phase: "stopped"
+```
+
+When you apply this manifest, the operator will:
+
+* Store the current `numberOfInstances` in `status.previousNumberOfInstances`
+* Store the connection pooler replica counts in `status.previousPoolerInstances`
+* Set `spec.numberOfInstances` to 0
+* Scale down the StatefulSet to 0 replicas
+* Scale down the connection pooler deployments to 0 replicas
+* Set `status.PostgresClusterStatus` to "Stopping", then "Stopped"
+
+### Waking up a Cluster
+
+To wake up a hibernated cluster, remove the `lifecycle.phase` field or set it to
+an empty value:
+
+```yaml
+apiVersion: "acid.zalan.do/v1"
+kind: postgresql
+metadata:
+  name: acid-test-cluster
+spec:
+  teamId: "test-team"
+  # ... other cluster parameters
+  # lifecycle.phase is not set or is removed
+```
+
+When you apply this manifest, the operator will:
+
+* Restore `numberOfInstances` from `status.previousNumberOfInstances`
+* Restore the connection pooler replica counts from `status.previousPoolerInstances`
+* Scale up the StatefulSet to the previous replica count
+* Scale up the connection pooler deployments to the previous replica counts
+* Set `status.PostgresClusterStatus` to "Updating", then "Running"
+* Clear `status.previousNumberOfInstances` and `status.previousPoolerInstances`
+
+### Cluster Status During Lifecycle Transitions
+
+| Status | Meaning |
+|--------|---------|
+| Running | Cluster is running normally |
+| Stopping | Cluster is transitioning to stopped state (pods terminating) |
+| Stopped | All pods have been terminated, cluster is hibernated |
+
+### Restrictions During Hibernate
+
+* **During Stopping**: All spec changes are blocked. You must wait for the cluster
+  to reach the Stopped state before making changes.
+
+* **During Stopped**: Spec changes are blocked unless you remove `lifecycle.phase`
+  to wake up the cluster. This prevents accidental modifications to a hibernated
+  cluster.
+
+### Connection Pooler Behavior
+
+The connection pooler is automatically scaled alongside the cluster:
+
+* When the cluster hibernates, the pooler is scaled to 0 replicas
+* When the cluster wakes up, the pooler is restored to its previous replica count
+* The previous replica counts are stored in `status.previousPoolerInstances`
+
+Note: If the connection pooler was already at 0 replicas before hibernate, it
+will remain at 0 after wake-up.
+
 ## Setting up a standby cluster
 
 Standby cluster is a [Patroni feature](https://github.com/zalando/patroni/blob/master/docs/replica_bootstrap.rst#standby-cluster)

manifests/postgresql.crd.yaml

@@ -3246,6 +3246,13 @@ spec:
                   - name
                   type: object
                 type: array
+              lifecycle:
+                description: LifecycleSpec describes the lifecycle state of a Postgres
+                  cluster.
+                properties:
+                  phase:
+                    type: string
+                type: object
               logicalBackupRetention:
                 type: string
               logicalBackupSchedule:
@@ -4197,6 +4204,14 @@ spec:
             properties:
               PostgresClusterStatus:
                 type: string
+              previousNumberOfInstances:
+                format: int32
+                type: integer
+              previousPoolerInstances:
+                type: object
+                additionalProperties:
+                  format: int32
+                  type: integer
             required:
             - PostgresClusterStatus
             type: object

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

@@ -9,6 +9,8 @@ const (
 	ClusterStatusSyncFailed   = "SyncFailed"
 	ClusterStatusAddFailed    = "CreateFailed"
 	ClusterStatusRunning      = "Running"
+	ClusterStatusStopping     = "Stopping"
+	ClusterStatusStopped      = "Stopped"
 	ClusterStatusInvalid      = "Invalid"
 )
 

pkg/apis/acid.zalan.do/v1/postgresql.crd.yaml

@@ -3246,6 +3246,13 @@ spec:
                   - name
                   type: object
                 type: array
+              lifecycle:
+                description: LifecycleSpec describes the lifecycle state of a Postgres
+                  cluster.
+                properties:
+                  phase:
+                    type: string
+                type: object
               logicalBackupRetention:
                 type: string
               logicalBackupSchedule:
@@ -4197,6 +4204,14 @@ spec:
             properties:
               PostgresClusterStatus:
                 type: string
+              previousNumberOfInstances:
+                format: int32
+                type: integer
+              previousPoolerInstances:
+                type: object
+                additionalProperties:
+                  format: int32
+                  type: integer
             required:
             - PostgresClusterStatus
             type: object

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

@@ -115,6 +115,7 @@ type PostgresSpec struct {
 	TLS                       *TLSDescription    `json:"tls,omitempty"`
 	AdditionalVolumes         []AdditionalVolume `json:"additionalVolumes,omitempty"`
 	Streams                   []Stream           `json:"streams,omitempty"`
+	Lifecycle                 *LifecycleSpec     `json:"lifecycle,omitempty"`
 	Env                       []v1.EnvVar        `json:"env,omitempty"`
 
 	// deprecated
@@ -257,6 +258,11 @@ type StandbyDescription struct {
 	StandbyPrimarySlotName string `json:"standby_primary_slot_name,omitempty"`
 }
 
+// LifecycleSpec describes the lifecycle state of a Postgres cluster.
+type LifecycleSpec struct {
+	Phase string `json:"phase,omitempty"`
+}
+
 // TLSDescription specs TLS properties
 type TLSDescription struct {
 	// +required
@@ -302,7 +308,9 @@ type UserFlags []string
 
 // PostgresStatus contains status of the PostgreSQL cluster (running, creation failed etc.)
 type PostgresStatus struct {
-	PostgresClusterStatus string `json:"PostgresClusterStatus"`
+	PostgresClusterStatus     string           `json:"PostgresClusterStatus"`
+	PreviousNumberOfInstances int32            `json:"previousNumberOfInstances,omitempty"`
+	PreviousPoolerInstances   map[string]int32 `json:"previousPoolerInstances,omitempty"`
 }
 
 // ConnectionPooler Options for connection pooler

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

@@ -101,6 +101,16 @@ func (postgresStatus PostgresStatus) Creating() bool {
 	return postgresStatus.PostgresClusterStatus == ClusterStatusCreating
 }
 
+// Stopping status of cluster
+func (postgresStatus PostgresStatus) Stopping() bool {
+	return postgresStatus.PostgresClusterStatus == ClusterStatusStopping
+}
+
+// Stopped status of cluster
+func (postgresStatus PostgresStatus) Stopped() bool {
+	return postgresStatus.PostgresClusterStatus == ClusterStatusStopped
+}
+
 func (postgresStatus PostgresStatus) String() string {
 	return postgresStatus.PostgresClusterStatus
 }