Dependencies
The OLIVE Platform relies on a set of third-party infrastructure components. The Cluster operator installs and maintains them as cluster infrastructure; as the Platform Administrator you validate that they are present and configure the OLIVE-specific parts of each one.
That second half is easy to underestimate. Several of these components need settings that only make sense if you know what OLIVE does with them — a Vault policy scoped to the paths tenant onboarding writes, a Keycloak issuer that matches what every service validates against. Installing the component is not the same as meeting OLIVE's requirements of it.
The exact versions of these dependencies are tied to the OLIVE Platform release version you are deploying.
A compatibility matrix listing the tested combinations of OLIVE Platform and third-party dependency versions is published per release. For the current release see 1.3.1 → Compatibility Matrix changes, which lists the rows that differ, and 1.3.0 → Compatibility Matrix for the full matrix. Share it with your cluster operator to ensure alignment.
For additional information about OLIVE Platform dependencies on cloud technology and the overall platform architecture, refer to the Platform Technology Overview documentation.
Vault, Keycloak, Traefik, the Istio mesh configuration and Kyverno each have a setup chart you may use. These charts are not part of the OLIVE platform — they are provided as-is, unsupported, to make setup easier. What is part of the platform is the documented set of requirements each component has to meet, which you can satisfy any way you like.
Infrastructure Charts covers both: what has to be true, and one way of getting there.
Development Deployment
For a development deployment, the following third-party infrastructure elements must be validated as installed and available in the cluster:
HashiCorp Vault
Used for managing connectivity API keys, database credentials, general secrets and sensitive configuration values during development.
- Provides secure injection of secrets into pods without hardcoding them in manifests.
- Cluster operator: Installation, the instance itself, seal/unseal and key custody
- Platform Administrator: The KV v2 mount, the Kubernetes auth method, and the bootstrap policy and role that tenant onboarding authenticates with — see Infrastructure Charts → HashiCorp Vault
- Documentation: HashiCorp Vault documentation
CloudNativePG (CNPG) operator
Required by the workflow engine to provision and manage the PostgreSQL database.
- Only relevant for OLIVE Platform versions ≥ 1.1.0; not relevant for 1.0.0.
- Installed once per Kubernetes cluster.
- Cluster operator: Installation and configuration
- Platform Administrator: Verify that CNPG is properly installed, using the commands below
- Documentation: CloudNativePG installation and upgrade
To verify CNPG installation, ask your cluster operator to confirm these commands succeed:
# Verify the operator deployment
kubectl -n cnpg-system get deploy,po
# Verify the custom resource definition
kubectl get crd | grep postgresql.cnpg.io
Traefik ingress controller
Serves every OLIVE route. Required in every deployment, including development.
- The only supported ingress controller from 1.3. OLIVE renders Traefik router annotations and a chain of
Middlewareresources, which have no equivalent for another controller; NGINX Ingress was deprecated in 1.2 and is not supported. - RKE2 provides Traefik by default. On EKS and other clusters it is installed separately — either way, OLIVE needs the
traefik.io/v1alpha1MiddlewareCRD, thewebandwebsecureentry points, and thekubernetesIngressandkubernetesCRDproviders, all of which are upstream defaults. - It must also carry the
redirectErrorsplugin, which is not built in. Without it an unauthenticated user meets an error page instead of a login page. - Cluster operator: installation, the plugin, the load-balancer in front of it, and
X-Forwarded-*trust - Platform Administrator:
global.ingress.className, andglobal.networkPolicy.ingressController— which must match the Traefik pods' real namespace and labels, because a selector matching nothing closes every route into the namespace - What the controller has to provide, in full: Cluster Operator → Ingress Requirements. Where Traefik is not already installed — typically EKS — there is an optional as-is chart for it,
traefik-infra, which also ships the required plugin
cert-manager
Issues the TLS certificates for the tenant hostnames and the Keycloak host.
- OLIVE renders the
Certificateobjects; theClusterIssuerthey reference is not created by OLIVE. - Cluster operator: installation, the
ClusterIssuer, its ACME account or CA key material, and the solver configuration - Platform Administrator:
global.clusterIssuer, and the hostnames the certificate has to cover - If you issue certificates another way, create the Secret yourself with the expected name and do not deploy the
certificatechart — OLIVE needs the Secret, not cert-manager. See Ingress Requirements → TLS - Documentation: cert-manager documentation
Production Deployment
For a production deployment, the following additional infrastructure components must be validated as installed and available:
Keycloak
Provides identity and access management (IAM) for the OLIVE Platform.
- Handles Single Sign-On (SSO), OIDC clients, user federation, and multi-tenant authentication.
- Cluster operator: Installation, the database, DNS, and a publicly trusted TLS certificate for the Keycloak host; multi-factor authentication for operator-level administrative identities
- Platform Administrator: Realm policy, MFA enforcement and user lifecycle; the admin credential in Vault; and the issuer matching what every OLIVE service validates against — see Infrastructure Charts → Keycloak
- Documentation: Keycloak documentation
HashiCorp Vault (production-grade setup)
Required for managing secrets and sensitive data in production environments.
- Must be deployed in HA mode with persistent backend storage.
- Integrated with Kubernetes via the Vault Secrets Operator — exactly one instance per cluster.
- Cluster operator: Installation, HA configuration, storage, seal/unseal and key custody
- Platform Administrator: The KV v2 mount, Kubernetes auth, the bootstrap policy and role, and the per-tenant roles applied during onboarding — see Infrastructure Charts → HashiCorp Vault
- Documentation: HashiCorp Vault documentation
Recommended GitOps Deployment
We recommend adopting a GitOps workflow for all environments.
Argo CD
Our validated and supported deployment tool.
- Automatically synchronizes desired state (Helm charts, manifests, values) from Git repositories into the Kubernetes cluster.
- Honours the sync waves the OLIVE charts declare, which is what enforces the correct installation order of the infrastructure Jobs, the stateless services and the tenant resources.
- Cluster operator: Installation, initial configuration, and the AppProject destinations the OLIVE deployments sync into
- Platform Administrator: Configuration for OLIVE Platform deployments and management of ApplicationSets
In production deployments, Argo CD is mandatory. Tenant onboarding and offboarding are driven from the infrastructure Git repository, and the sync waves the charts declare are what enforce the installation order — so consistency and repeatability depend on it.
Documentation: Argo CD documentation
Next Steps
After validating that all required dependencies are installed and properly configured, proceed to:
- Infrastructure Charts – What OLIVE requires of Vault, Keycloak, the service mesh and Kyverno, and the charts OLIVE provides to set them up.
- Chart Values – Review and prepare the
values.yamlconfiguration file for your specific deployment type (development or production) before installation.