Install Hybrid Manager on Red Hat OpenShift using EDB Postgres AI Operator

Install Operator dependencies

EDB Postgres AI Operator has an explicit dependency cert-manager operator for Red Hat OpenShift (RHOS), which manages all namespaces. It's installed automatically before EDB Postgres AI Platform Operator.

The operator for Red Hat also has implicit dependencies including:

  • RHOS Service Mesh 2, which manages all namespaces
  • Node Feature Disovery Operator, which manages the namespace openshift-nfd
  • NVIDIA GPU Operator, which manages the namespace nvidia-gpu-operator

If you've already installed these operators, you don't need to reinstall them.

Install the operator for RHOS

  1. To install EDB Postgres AI Operator on RHOS, first ensure you've prepared the image pull secret for accessing the EDB Cloudsmith repository. Also verify that you have the necessary permissions to pull images from the repository docker.enterprisedb.com/pgai-platform/edb-hcp-operator/manager.

    export DOCKER_SERVER="docker.enterprisedb.com"
    export DOCKER_USERNAME="<your-docker-user-name>"
    export DOCKER_PASSWORD="<your-docker-password>"
  2. Create a new OC project for EDB Postgres AI Operator:

oc new-project edb-hcp-operator-system
  1. Set up the image pull secret to access the EDB Cloudsmith repository:
oc create secret docker-registry edb-hcp-operator-pull-secret \
  --docker-server="$DOCKER_SERVER" \
  --docker-username="$DOCKER_USERNAME" \
  --docker-password="$DOCKER_PASSWORD" \
  --namespace-edb-hcp-operator-system
  1. Use OpenShift Container Platform (OCP) to install EDB Postgres AI Operator for Red Hat OpenShift:

  2. Log in to the OCP web console.

  3. Select Operators > OperatorHub.

  4. In the filter box, enter EDB Postgres AI.

  5. Select EDB Postgres AI Operator and select Install.

  6. On the Install Operator page: a. For the operator, select Installed Namespace using the edb-hcp-operator-system name you created earlier. b. Select Install.

Tip

On RHOS, cert-manager is a required dependency for Hybrid Manager. When installing cert-manager, so that it works correctly, ensure that Installation Mode is configured as All namespaces on the cluster.

Install Hybrid Manager on RHOS

After EDB Postgres AI Operator is installed on your RHOS cluster, you need to use the Operator to install Hybrid Manager. The following YAML file initiates the Hybrid Manager installation:

apiVersion: edbpgai.edb.com/v1alpha1
kind: HybridControlPlane
metadata:
  name: edbpgai
spec:
  source:
    useLocalKustomizations: true
  flavour: rhos
  version: v1.2.0+appl
  imageRegistry: ${CONTAINER_REGISTRY_URI}
  imagePullSecrets:
    - name: edb-cred
      namespace: edbpgai-bootstrap
  globalParameters:
    internal_backup_folder: 4befab698c8d
    portal_domain_name: ${PORTAL_DOMAIN_NAME}
    default_ingress_domain: ${DEFAULT_INGRESS_DOMAIN}
    storage_class: ${STORAGE_CLASS}
  componentsParameters:
    edb-migration-portal:
      enable_pdb: "false"
    transporter-data-operator:
      fips_enabled: "${TRANSPORTER_FIPS_ENABLED}"
    transporter-dp-agent:
      rw_service_url: https://${TRANSPORTER_RW_SERVICE_DOMAIN_NAME}/transporter
    transporter-rw-service:
      domain_name: ${TRANSPORTER_RW_SERVICE_DOMAIN_NAME}
    upm-api-pnxt-metric:
      log_level: info
      tsdb_host: thanos-query.monitoring.svc.cluster.local
      tsdb_port: "9090"
    upm-api-pnxt-pd:
      log_level: info
    upm-beaco-ff-base:
      usage_generator_interval: 24h
    upm-beaco-usages:
      usage_generator_interval: 24h
    upm-beacon:
      server_host: ${BEACON_SERVICE_DOMAIN_NAME}
      beacon_server_replicas: "1"
      external_metrics_replicas: "1"
      beacon_db_hibernation: "\"off\""
    upm-istio-gateway:
      openshift_console_domain_name: ${OPENSHIFT_CONSOLE_DOMAIN_NAME}
      cookie_aeskey: ${AES_256_KEY}
      ingress_http_node_port: "32542"
      ingress_https_node_port: "30288"
      ingress_grpc_tls_node_port: "30290"
    upm-prometheus:
      memory: 4Gi
    upm-trust-manager:
      cert_manager_namespace: ${CERT_MANAGER_NAMESPACE}
  pgai:
    portal:
      authentication:
        clientSecret: edb-postgres-ai-secret
        idpConnectors: []
        staticPasswords:
          - email: ${AUTHENTICATION_EMAIL}
            hash: ${AUTHENTICATION_HASH}
            username: ${AUTHENTICATION_USERNAME}
            userID: ${AUTHENTICATION_USER_ID}
  beaconAgent:
    applianceProviderEnabled: true
    plaintext: true
    beaconServerAddress: "upm-beacon-server-ingestion.upm-beacon.svc.cluster.local:8080"
    tlsInsecure: false
    location: "devspatcher"
    logLevel: "debug"
    beaconDBConnected: true
    inCluster: true
    provisioning:
      loadBalancersEnabled: true
      nodePortDomain: "nodeportdomain.com"
      imageDiscovery: true
      imagesetDiscoveryAuthenticationType: "basic"
      imagesetDiscoveryContainerRegistryURL: "us-east1-docker.pkg.dev"
    providers:
      - appliance

After the YAML has been applied, you can use the following command to monitor the installation status:

oc get hybridcontrolplane edbpgai -o yaml

Three different statuses can be returned, depending on the phase the installation is currently in:

- deployed # the installation was successful
- deploying # the installation is still in progress
- failed: # the install was unsuccessful

After the install is successful, updating DNS zone records and verify the install**.

Uninstalling Hybrid Manager installed via Operator

Warning

The following procedure uninstalls Hybrid Manager. Use this procedure only if you're sure you want to uninstall Hybrid Manager.

There may be cases where you want to delete Hybrid Manager. Because Hybrid Manager is protected from deletion by default, you must first disable that protection:

kubectl annotate hybridcontrolplane edbpgai biganimal.enterprisedb.io/deletion-protect=disabled --overwrite

You can then uninstall:

kubectl delete hybridcontrolplane edbpgai --wait=false

To monitor the process, run:

kubectl get hybridcontrolplane edbpgai -o yaml

Could this page be better? Report a problem or suggest an addition!