Resize PGO cluster storage volumes and compute limits

As workload demands change, you will need to adjust the resources allocated to your PostgreSQL cluster. PGO supports two classes of resize: storage volume expansion and compute (CPU and memory) changes. Each behaves differently in terms of disruption to your running cluster.

Resize storage volumes

PostgreSQL instance storage is defined by dataVolumeClaimSpec (and optionally walVolumeClaimSpec) inside each instances entry. To increase the volume size, update the resources.requests.storage value and apply the change.

spec:
  instances:
    - name: instance1
      dataVolumeClaimSpec:
        accessModes:
          - ReadWriteOnce
        resources:
          requests:
            storage: 10Gi  # increased from 5Gi

PVC expansion requires that the underlying StorageClass has allowVolumeExpansion: true. Shrinking a volume is not supported by Kubernetes and will be rejected.

PGO propagates the updated storage request to the underlying PersistentVolumeClaim. The storage expansion is performed by the Kubernetes volume provisioner and typically does not require a pod restart, though this depends on the provisioner. You can watch the resize progress by checking the PersistentVolumeResizing condition on the PostgresCluster status:

kubectl get postgrescluster hippo -o jsonpath='{.status.conditions}'

WAL volume

If you configured a separate WAL volume using walVolumeClaimSpec, resize it the same way:

spec:
  instances:
    - name: instance1
      walVolumeClaimSpec:
        accessModes:
          - ReadWriteOnce
        resources:
          requests:
            storage: 5Gi

Automatic volume growth

PGO includes an AutoGrowVolumes feature gate (alpha, enabled by default) that can automatically expand volumes before they run out of space, eliminating the need for manual intervention. To configure auto-grow, add an autoGrow stanza inside dataVolumeClaimSpec or walVolumeClaimSpec. Auto-grow requires that the PVC has a storage limit set.

spec:
  instances:
    - name: instance1
      dataVolumeClaimSpec:
        accessModes:
          - ReadWriteOnce
        resources:
          requests:
            storage: 10Gi
          limits:
            storage: 50Gi
        autoGrow:
          trigger: 75
          maxGrow: 20Gi

AutoGrowSpec fields

Field	Type	Default	Description
`trigger`	integer	`75`	Percentage of used space that triggers expansion. Valid range: 50–90.
`maxGrow`	resource.Quantity	—	Maximum size to which the volume can be automatically expanded. If not set, the volume grows by 50% of the original size each time the trigger is exceeded.

The AutoGrowVolumes feature gate is alpha-stage but enabled by default. If you need to disable it, set AutoGrowVolumes=false in the operator’s feature gate configuration. See Feature Gates.

Resize compute (CPU and memory)

CPU and memory for PostgreSQL instance pods are controlled by the resources field inside each instances entry. This follows the standard Kubernetes ResourceRequirements structure.

spec:
  instances:
    - name: instance1
      resources:
        requests:
          cpu: 500m
          memory: 256Mi
        limits:
          cpu: "2"
          memory: 1Gi

Changing the resources field causes the instance pods to restart. PGO performs a rolling restart, so the cluster will remain available as long as you have replicas running. Set minAvailable to control how many pods must stay up during the rollout.

Control disruption with minAvailable

Use minAvailable to ensure a minimum number of pods remain healthy during a rolling restart:

spec:
  instances:
    - name: instance1
      replicas: 3
      minAvailable: 2
      resources:
        requests:
          cpu: "1"
          memory: 512Mi

Changes that do not restart pods

The following storage-related changes do not cause pod restarts:

Increasing dataVolumeClaimSpec.resources.requests.storage (volume expansion only, no pod restart when the StorageClass supports online resizing)
Modifying autoGrow.trigger or autoGrow.maxGrow

The following changes do cause pod restarts:

Updating resources (CPU/memory) on an instance set
Changing dataVolumeClaimSpec.storageClassName
Adding or removing a walVolumeClaimSpec
Changing affinity, tolerations, or topologySpreadConstraints

Next steps

Upgrade PostgreSQL versions

Perform minor and major version upgrades on your cluster.

Customize cluster configuration

Tune postgresql.conf parameters and pod scheduling constraints.

Get Started

Installation

Core Operations

Day-Two Operations

Security

Resize PGO cluster storage volumes and compute limits

Resize storage volumes

WAL volume

Automatic volume growth

AutoGrowSpec fields

Resize compute (CPU and memory)

Control disruption with minAvailable

Changes that do not restart pods

Next steps

Upgrade PostgreSQL versions

Customize cluster configuration

​Resize storage volumes

​WAL volume

​Automatic volume growth

​AutoGrowSpec fields

​Resize compute (CPU and memory)

​Control disruption with minAvailable

​Changes that do not restart pods

​Next steps

Upgrade PostgreSQL versions

Customize cluster configuration

Resize storage volumes

WAL volume

Automatic volume growth

AutoGrowSpec fields

Resize compute (CPU and memory)

Control disruption with minAvailable

Changes that do not restart pods

Next steps