Upgrading from 2026.3 to the 2026.4 Innovation Release Innovation Release

Upgrade from 2026.3 to 2026.4.

Role: Infrastructure engineer

Prerequisites

• Administrative access to the Kubernetes cluster (kubectl, helm).

• Access to the HM installation configuration file or values.yaml file used for the current installation. To save the contents of your current installation configuration:

  → WrapCopy

  helm get values edbpgai-bootstrap -n edbpgai-bootstrap > values.yaml

• Required tools:

  • helm (Latest version recommended)
  • kubectl
  • edbctl version 1.4.0 or later (or a build from main)

Upgrade 2026.3 → 2026.4

This release introduces the dbaas scenario, requires new secrets for upm-system-db, and includes mandatory pre-upgrade steps to prevent data corruption and resource conflicts.

Create new secrets for upm-system-db

Before upgrading, create the new secrets required by the upm-system-db component:

edbctl setup create-install-secrets --version v2026.4.0

Delete fluent-bit resources

Label changes in 2026.4 make the existing fluent-bit StatefulSet and DaemonSet immutable. Delete both resources before upgrading to prevent the upgrade from failing:

kubectl delete statefulset fluent-bit-k8s-events -n edb-observability
kubectl delete daemonset fluent-bit -n edb-observability

The operator recreates them automatically during the upgrade.

Delete the statistics collector database

1. Delete the cluster:

   → WrapCopy

   kubectl -n upm-api-stats-collector delete cluster stats-collector-db

   If the cluster doesn't terminate promptly, force-delete the pods:

   → WrapCopy

   kubectl delete pod --force -n upm-api-stats-collector -l cnpg.io/cluster=stats-collector-db

2. Verify all associated resources are removed before proceeding:

   → WrapCopy

   kubectl -n upm-api-stats-collector get cluster stats-collector-db
   kubectl -n upm-api-stats-collector get pod stats-collector-db-1
   kubectl -n upm-api-stats-collector get pvc stats-collector-db-1

   Each command returns NotFound when the resource is gone. Proceed only once all three return NotFound.

Perform the upgrade

For Helm-based installations

1. Update your Helm repository:

   → WrapCopy

   helm repo update

2. If your SKU permits, add dbaas to the scenarios field in your values.yaml:

   → WrapCopy

   scenarios: "core,migration,ai,analytics,dbaas"

3. Perform the upgrade:

   → WrapCopy

   helm upgrade --install edbpgai-bootstrap enterprisedb-edbpgai/edbpgai-bootstrap \
     --version 2026.4.0 \
     --values values.yaml \
     --namespace edbpgai-bootstrap \
     --wait

For operator-based (CRD) installations

The dbaas scenario wasn't present in the 2026.3 CRD schema and can't be added while still running the 2026.3 operator. Upgrade the operator first, then update the HybridControlPlane resource.

1. Upgrade the edb-hcp-operator Helm chart to version 2026.4.0:

   → WrapCopy

   helm template \
     --namespace edbpgai-bootstrap \
     --install \
     --version 2026.4.0 \
     --include-crds \
     edb-hcp-operator enterprisedb-edbpgai/edb-hcp-operator \
     --set "controllerManager.manager.image.repository=docker.enterprisedb.com/pgai-platform/edb-hcp-operator/manager" \
     --set "controllerManager.manager.image.tag=2026.4.0" \
     --set "imagePullSecrets[0].name=edb-cred" \
     | kubectl apply -f -

2. Update the version and, f your SKU permits, add the dbaas scenario in a single patch command:

   → WrapCopy

   kubectl patch hcp edbpgai --type json \
     --patch '[{"op":"add", "path":"/spec/scenarios/-", "value":"dbaas"}, {"op":"replace", "path":"/spec/version", "value":"v2026.4.0"}]'

   If you aren't enabling dbaas, update only the version:

   → WrapCopy

   kubectl patch hcp edbpgai --type json \
     --patch '[{"op":"replace", "path":"/spec/version", "value":"v2026.4.0"}]'

   See The dbaas scenario for details on what happens if dbaas isn't enabled.

3. Trigger the upgrade by annotating the resource:

   → WrapCopy

   kubectl annotate hybridcontrolplane edbpgai --overwrite edbpgai.com/ready-for-upgrade=true

4. The operator detects the version change and begins the rolling update. Monitor the progress with:

   → WrapCopy

   kubectl get hybridcontrolplane edbpgai -w

Enable beacon server usage telemetry

2026.4 introduces a new boolean field for enabling usage server telemetry collection. To enable it, set enableEDBUsageReporting to true alongside the existing required fields.

For operator-based installations, update your HybridControlPlane CR:

spec:
  beaconServer:
    enableEDBUsageReporting: true
    hybridManagerInstallation: "<your-installation-name>"
  globalParameters:
    company_code: "<your-company-code>"

For Helm-based installations, add the equivalent values to your values.yaml:

beaconServer:
  enableEDBUsageReporting: true
  hybridManagerInstallation: "<your-installation-name>"
parameters:
  global:
    company_code: "<your-company-code>"

After updating your configuration, re-annotate to apply the changes:

kubectl annotate hybridcontrolplane edbpgai --overwrite edbpgai.com/ready-for-upgrade=true

The dbaas scenario

The dbaas scenario is new in 2026.4 and is separate from core. It enables database cluster management — including cluster provisioning, backups, storage locations, and templates. For an overview of all available scenarios, see Planning your architecture.

If dbaas isn't included during the upgrade:

• Cluster management APIs return 404.
• The Clusters section is hidden in the HM console.
• Existing Postgres clusters continue running but are no longer manageable through the platform.

You can add dbaas at any time after the upgrade by patching the CR and re-triggering reconciliation:

kubectl patch hcp edbpgai --type json \
  -p '[{"op":"add", "path":"/spec/scenarios/-", "value":"dbaas"}]'

kubectl annotate hybridcontrolplane edbpgai \
  --overwrite edbpgai.com/ready-for-upgrade=true

Troubleshooting

edb-migration-copilot fails to install

If the upgrade fails with an error similar to:

failed to execute kapp command, error: update secret/ragchew-ai-vendor-secrets (v1) namespace: edb-migration-copilot

Delete the conflicting secret and re-trigger the upgrade:

kubectl delete secret -n edb-migration-copilot ragchew-ai-vendor-secrets
kubectl annotate hybridcontrolplane edbpgai --overwrite edbpgai.com/ready-for-upgrade=true