keenPG on Kubernetes · KPK

Write what you want.
The operator keeps it true.

keenPG on Kubernetes runs PostgreSQL clusters natively on OpenShift, Kubernetes, OKD and Rancher. Declare a cluster in YAML and its operator makes it real, then keeps it that way: replacing failed pods, promoting replicas, taking backups and rolling out upgrades.
1YAML file
0manual failovers
100%open-source PostgreSQL
orders-cluster.yaml0/3 running
apiVersion: keenpg.io/v1
kind: PostgresCluster
metadata:
  name: orders
spec:
  postgresVersion: 17
  instances: 3
  backups:
    schedule: "0 */6 * * *"
    retention: 30d
  monitoring: enabled
# that's the whole cluster
namespace: orders0 pods
change instances · click any pod to delete it
How the operator works

From YAML to a
running cluster.

  1. 1You declare what you wantA few lines of YAML: a cluster, three instances, backups on.
  2. 2The operator reads itIt watches your cluster definitions and works out what has to exist.
  3. 3Pods appearA primary and two replicas start, with storage, services and backups wired up.
  4. 4It keeps it that wayIf a pod dies, the operator replaces it. Desired state, always.
orders.yamlapiVersion: keenpg.io/v1kind: PostgresClustermetadata: name: ordersspec: instances: 3 backups: enabledoperatororders-0 · primaryorders-1orders-2orders-2 · lostorders-3 · createddesired = actual ✔
What the operator does for you

Day-two operations,
automated.

Self-healing

Failed pods are replaced and replicas promoted without anyone being paged.

Backups to any bucket

Scheduled full and incremental backups to S3, GCS or Azure Blob, with point-in-time restore.

Rolling upgrades

Minor and major versions roll out one pod at a time, with replicas first and the primary last.

Connection pooling

A built-in PgBouncer pool keeps thousands of app connections from overwhelming Postgres.

Metrics out of the box

Prometheus exporters and Grafana dashboards are deployed alongside every cluster.

Secure by default

TLS everywhere, rotated credentials in Kubernetes Secrets and pod security standards.

GitOps-native

From commit to cluster.

Your database lives in Git like the rest of your stack. Review it, version it, roll it back.

Commitinstances: 3 → 4
Review & mergeArgo CD · Flux
Operator reconcilesdesired = actual
Cluster updatedzero downtime
Red Hat OpenShift (OCP)KubernetesOKDRancherAmazon EKSAzure AKSGoogle GKE

One helm install
away from calm.

Add keenPG Experts when you want a team behind it.