SupportConsoleDevelopersStart a trialContact

Chapter 2. Enable MCP server lifecycle management

As a cluster administrator, you can enable the MCP Lifecycle Operator to make MCP server deployment available to your team. After you enable the operator and the mcpCatalog dashboard feature flag, developers can deploy and manage MCP servers directly from the AI Hub MCP Catalog.

Important

MCP Lifecycle Operator is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

2.1. MCP Lifecycle Operator
Copy link

Important

MCP Lifecycle Operator is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

The MCP Lifecycle Operator is a component in Red Hat OpenShift AI that manages the deployment and lifecycle of Model Context Protocol (MCP) servers. When enabled, it provides the runtime backend so that developers can deploy MCP servers directly from the AI Hub MCP Catalog without writing Kubernetes manifests or sourcing container images manually.

The MCP Lifecycle Operator is the cluster-level controller managed by the Red Hat OpenShift AI Operator through the DataScienceCluster custom resource. It watches for MCPServer custom resources and coordinates the overall deployment lifecycle by automatically provisioning the required Deployments, Services, NetworkPolicies, and cluster-internal URLs for service discovery.

2.1.1. Component controller pattern
Copy link

The MCP Lifecycle Operator follows the same component controller pattern as other optional Red Hat OpenShift AI components such as MLflowOperator, FeastOperator, and Open GenAI Stack (OGX). You manage the operator by setting the managementState field in the DataScienceCluster custom resource.

The mcplifecycleoperator component defaults to Removed in the DataScienceCluster for Red Hat OpenShift AI 3.5. You must explicitly set managementState to Managed to enable MCP server deployment on your cluster.

2.1.2. Relationship to the AI Hub MCP Catalog
Copy link

The AI Hub MCP Catalog provides the discovery and deployment interface for MCP servers. The catalog UI is a federated plugin in the Red Hat OpenShift AI dashboard and is pre-loaded with MCP servers from Red Hat, technology partners, and the open source community.

The MCP Lifecycle Operator provides the deployment runtime backend for the catalog. Without the operator enabled, the deploy action in the catalog UI is unavailable. After you enable the operator, developers in your team can browse the catalog and deploy MCP servers to their namespaces without additional configuration.

A separate model-metadata-collection data container provides the catalog metadata for MCP server discovery.

2.1.3. Relationship to the MCP Gateway Operator
Copy link

The MCP Gateway Operator is a separate component that provides routing and security for MCP servers at the gateway layer. It is not part of the Red Hat OpenShift AI Operator distribution and is not automatically installed when you enable the MCP Lifecycle Operator.

If your MCP server deployment scenarios require centralized routing or access control through the MCP Gateway, install the MCP Gateway Operator separately before enabling MCP server deployments.

Note

Not all MCP server deployment scenarios require the MCP Gateway Operator.

2.1.4. Distinction from the RHOAI MCP server
Copy link

The MCP Lifecycle Operator is a Kubernetes operator for running MCP servers as native workloads on OpenShift. It is not itself an MCP server.

rhoai-mcp
An MCP server that enables MCP-compatible AI coding agents, such as Claude Code, to interact with Red Hat OpenShift AI itself. It provides tools for model recommendation, project management, and Kubernetes manifest generation.
MCP Lifecycle Operator
A Kubernetes operator that deploys and manages MCP servers as native workloads on OpenShift.
Important

MCP Lifecycle Operator is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

The MCP Lifecycle Operator in Red Hat OpenShift AI 3.5 has the following limitations.

LimitationDetails

No backward compatibility for the Custom Resource Definition (CRD) API

The MCPServer custom resource API version is v1alpha1. Red Hat makes no backwards compatibility commitment for this release. CRD types and fields can change in a future release without a migration path.

Single-cluster deployment only

The operator manages MCP servers within a single Red Hat OpenShift AI cluster. Multi-cluster MCP server management is not validated.

No governance or policy enforcement

Rate limiting, access control, and policy enforcement for MCP servers are not included.

Dashboard UI requires feature flag

MCP lifecycle features in the Red Hat OpenShift AI dashboard are gated by the mcpCatalog feature flag in the OdhDashboardConfig custom resource. A cluster administrator must enable the flag before the MCP Catalog deploy action is available to users.

MCP Gateway Operator is a separate installation

The MCP Gateway Operator, which provides centralized routing and access control for MCP servers, is not automatically installed when you enable the MCP Lifecycle Operator. If your deployment requires MCP server routing through the gateway, you must install the MCP Gateway Operator separately.

Disconnected environment image mirroring

The MCP Lifecycle Operator image is included in the Red Hat OpenShift AI operator catalog relatedImages for OLM-based air-gapped mirroring and requires no extra mirroring steps. However, individual MCP server images available in the catalog might require additional mirroring steps in disconnected environments. Consult each server’s metadata for image references.

2.3. Enable the MCP Lifecycle Operator
Copy link

Important

MCP Lifecycle Operator is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

Enable the MCP Lifecycle Operator so that developers on your team can deploy MCP servers from the AI Hub MCP Catalog. When enabled, the operator manages the full lifecycle of MCP server deployments in your cluster.

Prerequisites

  • Red Hat OpenShift AI 3.5 is installed on an OpenShift 4.22 or later cluster.
  • You have cluster administrator privileges.
  • You have installed the OpenShift CLI (oc).

Procedure

  1. Identify the name of your DataScienceCluster object: 

    $ oc get datasciencecluster

  2. Enable the MCP Lifecycle Operator by patching the DataScienceCluster object: 

    $ oc patch datasciencecluster default-dsc --type=merge \
  -p '{"spec":{"components":{"mcplifecycleoperator":{"managementState":"Managed"}}}}'

    Replace default-dsc with the name of your DataScienceCluster object if different.

    Note

    You can also enable the MCP Lifecycle Operator by using the OpenShift web console. For more information, see Installing and uninstalling OpenShift AI.

  3. Enable the MCP lifecycle feature in the dashboard by patching the OdhDashboardConfig custom resource: 

    $ oc patch odhdashboardconfig odh-dashboard-config -n redhat-ods-applications \
  --type=merge -p '{"spec":{"dashboardConfig":{"mcpCatalog":true}}}'

Verification

  • Confirm that the MCP Lifecycle Operator pod is running: 

    $ oc get pods -n redhat-ods-applications -l app.kubernetes.io/name=mcp-lifecycle-operator

    The output shows at least one pod in Running status, for example: 

    NAME                                    READY   STATUS    RESTARTS   AGE
mcp-lifecycle-operator-<id>            1/1     Running   0          2m

  • Confirm that the mcplifecycleoperator component is reported as Managed in the DataScienceCluster object: 

    $ oc get datasciencecluster default-dsc \
  -o jsonpath='{.status.installedComponents.mcplifecycleoperator}'

    The expected output is true.

Next steps

2.4. Disable the MCP Lifecycle Operator
Copy link

Important

MCP Lifecycle Operator is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

Disable the MCP Lifecycle Operator when you no longer need MCP server lifecycle management on your cluster, or when you need to reclaim resources.

Important

Disabling the MCP Lifecycle Operator removes the operator and all resources it manages. MCP servers and their associated MCPServer custom resources might be cleaned up automatically. To confirm cleanup, verify that no MCPServer resources remain in user namespaces after disabling the operator. Developers can no longer deploy MCP servers from the AI Hub MCP Catalog until you re-enable the operator. The Deploy button in the MCP server details page is also disabled.

Prerequisites

  • You have cluster administrator privileges.
  • You have installed the OpenShift CLI (oc).
  • The MCP Lifecycle Operator is currently enabled.

Procedure

  1. Disable the MCP Lifecycle Operator by patching the DataScienceCluster object: 

    $ oc patch datasciencecluster default-dsc --type=merge \
  -p '{"spec":{"components":{"mcplifecycleoperator":{"managementState":"Removed"}}}}'

    Replace default-dsc with the name of your DataScienceCluster object if different.

  2. Optional: Disable the MCP lifecycle dashboard feature flag to hide the MCP catalog UI from developers: 

    $ oc patch odhdashboardconfig odh-dashboard-config -n redhat-ods-applications \
  --type=merge -p '{"spec":{"dashboardConfig":{"mcpCatalog":false}}}'

Verification

  • Confirm that no MCP Lifecycle Operator pods are running: 

    $ oc get pods -n redhat-ods-applications -l app.kubernetes.io/name=mcp-lifecycle-operator

    The command returns no output when the operator has been successfully removed.

  • Confirm that the mcplifecycleoperator component is reported as Removed in the DataScienceCluster object: 

    $ oc get datasciencecluster default-dsc \
  -o jsonpath='{.status.installedComponents.mcplifecycleoperator}'

    The expected output is false. If the field is missing from the output, the component is not installed.

  • Confirm that no MCPServer resources remain in user namespaces: 

    $ oc get mcpservers --all-namespaces

    The expected output is No resources found or an empty list.

Red Hat logoGithubRedditYoutubeTwitter

Learn

Try, buy, & sell

Communities

About Red Hat

We deliver hardened solutions that make it easier for enterprises to work across platforms and environments, from the core datacenter to the network edge.

Making open source more inclusive

Red Hat is committed to replacing problematic language in our code, documentation, and web properties. For more details, see the Red Hat Blog.

About Red Hat Documentation

We help Red Hat users innovate and achieve their goals with our products and services with content they can trust. Explore our recent updates.

Legal Notice

Theme

© 2026 Red HatBack to top