Deploy and configure Red Hat OpenShift Lightspeed with IBM watsonx Runtime

Red Hat OpenShift Lightspeed is a generative AI-based virtual assistant built into the OpenShift web console. Developers and administrators can use it to:

• Ask questions about OpenShift and its deployed components.

• Get real-time, context-aware answers from the cluster.

Backed by Red Hat’s expertise in OpenShift and mission-critical applications, Lightspeed helps users build skills faster, navigate the console more easily, and improve productivity in tasks such as troubleshooting and investigating cluster resources.

Lightspeed supports queries on:

• OpenShift Container Platform

• Kubernetes

• OpenShift Virtualization

• OpenShift Pipelines

• OpenShift Service Mesh

It does not respond to topics beyond this scope. For a full list of supported topics, refer to the latest documentation.

Lightspeed can connect to large language models (LLMs) from:

• IBM watsonx.ai

• Microsoft Azure OpenAI

• OpenAI

This flexibility helps organizations adopt Lightspeed across platforms while making use of their existing tech investments.

As of June 2025, Lightspeed is generally available. For more details, see the Red Hat announcement in the References and resources section.

Because it is context-aware, Lightspeed maintains conversation threads. You can reset the context by:

• Clearing chat history

• Reloading the console

• Restarting the Lightspeed operator

OpenShift Lightspeed supports several key use cases:

• Querying: Ask stand-alone questions about cluster resources and related follow-up questions within the same context.

• Knowledge finding: Get answers based on existing knowledge sources, including YAML definitions and other attachable resources.

• Troubleshooting: Investigate alerts with context pulled from the cluster.

• Logging: Keep or clear conversation history as needed.

Lightspeed also captures user feedback using thumbs up/down buttons to help improve its performance over time.

OpenShift Lightspeed supports the following OpenShift resource types:

Prerequisites

To complete this tutorial, you need:

• An IBM Cloud account.

• A Red Hat OpenShift cluster set up on IBM Cloud (bare metal or virtual server).

Procedure: Summary of steps

Follow these steps to complete the tutorial. Each step is explained in detail in the Detailed walkthrough section.

1. Set up object storage.

   • Create an IBM Cloud Object Storage instance

   • Create a bucket in the instance

2. Set up the IBM Cloud watsonx.ai Runtime.

3. Deploy the OpenShift Lightspeed Operator.

4. Configure the watsonx.ai Runtime.

   • Note the watsonx.ai URL for use when configuring the Lightspeed operator

5. Create OpenShift Lightspeed resources.

   • Create an API key

   • Create a project

   • Deploy the virtual assistant

   • Connect to the backend LLM

   • Create a secret

   • Create the OLSConfig custom resource

   • Verify that OpenShift Lightspeed is running

6. Access the Lightspeed Assistant.

7. Run natural language queries.

   • Ask a general stand-alone question

   • Follow up with related questions

   • Ask a question with attached cluster resources

   • Troubleshoot alerts

   • Start a new conversation after clearing previous history

8. Check infrastructure consumption from:

   • OpenShift (Developer and Administrator views)

   • IBM watsonx.ai Project Overview

Procedure: Detailed walkthrough

Step 1. Create IBM Cloud object storage and bucket

In this step, you will create an IBM Cloud object storage (COS) instance and a bucket.

• Region: Canada – Toronto

• Plan: Standard

• Resource Group: cloud-techsales

• Storage Class: Smart Tier (Regional)

Create COS Instance

1. From the IBM Cloud console, go to Navigation Menu → Infrastructure → Storage → Object Storage.

2. Click Create instance.

3. Fill in the following:

   • Pricing plan: Standard

   • Service name: <YOUR-SERVICE-INSTANCE-NAME>

   • Tags: <YOUR-SERVICE-INSTANCE-TAG>

   • Resource group: <YOUR-RESOURCE-GROUP>

4. Click Create.

Create Bucket

1. Go to Navigation Menu → Infrastructure → Storage → Object Storage → Instances.

2. Select your instance: <YOUR-SERVICE-INSTANCE-NAME>

3. Click Create bucket → Create a custom bucket.

4. Fill in the following:

   • Bucket name: <YOUR-CUSTOM-BUCKET-NAME> (must be unique)

   • Resiliency: Regional

   • Location: Canada – Toronto (ca-tor)

   • Storage class: Smart Tier

5. Click Create bucket.

   

Step 2. Create an IBM Cloud watsonx.ai Runtime instance

In this step, you will create a watsonx.ai Runtime instance in IBM Cloud.

• Region: Canada – Toronto

• Plan: Lite

• Resource Group: cloud-techsales

Follow these steps:

1. Go to Catalog → watsonx.ai Runtime.

2. Set the following options:

   • Location: Toronto (ca-tor)

   • Pricing plan: Lite

   • Service name: <YOUR-SERVICE-INSTANCE-NAME>

   • Tags: <YOUR-SERVICE-INSTANCE-TAG>

   • Resource group: <YOUR-RESOURCE-GROUP>

3. Accept the license agreement.

4. Click Create.

   

   After the Runtime instance is created, the IBM watsonx welcome screen appears.

   

   You can now see the following two resources listed in your IBM Cloud account under Resource list.

   

Step 3. Deploy the OpenShift Lightspeed operator from OperatorHub

Log in to the Red Hat OpenShift web console as an Administrator (with cluster admin permissions). From the Administrator view:

1. Go to Operators → OperatorHub.

2. Search for OpenShift Lightspeed Operator.

3. Install it using the default options.

   After installation, switch to the OpenShift-Lightspeed project. You will see the operator dashboard. At the time of writing, the operator version is 0.3.4.

   

   This screen shows the current configuration details of the OpenShift Lightspeed operator before setting up the OLSConfig custom resource.

   

Step 4. Configure watsonx.ai Runtime service

To configure the watsonx.ai Runtime service, start by creating an API key for the backend large language model.

Make sure to save the generated API key, you will need it later.

Steps:

1. Go to Create API key.

   • Name: <YOUR-API-KEY>

   • Description (optional): <YOUR-API-KEY-DESC>

2. Click Create.

   

   

   

3. Create a project where the backend large language model will be hosted. Make sure to save the Project ID for later use.

   Steps:

   a. Go to Create Project.

   b. Define details:

   • Name: <YOUR-PROJECT>

   • Description (optional): <YOUR-PROJECT-DESC>

   • Tags (optional): <YOUR-PROJ-TAG>

   c. Define storage

   • Select storage service: <YOUR-SERVICE-INSTANCE-NAME>

   d. Click Create

   

   After creating the project, you will be directed to the project overview screen.

   

4. Associate your project with a watsonx.ai Runtime instance. You will see a screen guiding you through this step. The large language model used here is granite-3-8b-instruct.

   

   Connect your project to the Runtime instance to host the large language model (LLM).

   

   Confirm that your Runtime instance is successfully associated with your project, as shown.

   

5. Create a deployment space to host the backend large language model. Make sure to save the Deployment Space GUID for later configuration.

   a. Create a deployment space.

   • Name: <YOUR-DEPLOYMENT-SPACE-NAME>

   • Description (Optional): <YOUR-DEPLOYMENT-SPACE-DESC>

   • Deployment Stage: Development

   • Tags (Optional): <YOUR-DEPLOYMENT-SPACE-TAG>

   Select Services:

   • Storage Service: <YOUR-SERVICE-INSTANCE-NAME>

   • watsonx.ai Runtime Service (Optional): `

   b. Click Create.

   

   

   

   

The following screen shows a fully configured watsonx.ai setup with the granite-3-8b-instruct large language model.

Step 5. Create OpenShift Lightspeed resources in OpenShift

In this step, you will create two YAML files:

• A Secret resource that stores the API key to connect OpenShift Lightspeed to the backend LLM.

• An OLSConfig custom resource that configures OpenShift Lightspeed using the watsonx.ai URL, project ID, and model name for the Toronto region.

After creating the YAML files, apply them using the OpenShift CLI (oc).

bvperepa:~$ cat ./<YOUR-LLM-PROVIDER-NAME>-llm-secret.yaml
apiVersion: v1
kind: Secret
metadata:
  name: credentials
  namespace: openshift-lightspeed
type: Opaque
stringData:
  apitoken: <watsonx.ai API key>

bvperepa:~$ cat ./<YOUR-LLM-PROVIDER-NAME>-llm-config-cr.yaml
apiVersion: ols.openshift.io/v1alpha1
kind: OLSConfig
metadata:
  name: cluster
spec:
  llm:
    providers:
      - name: <YOUR-LLM-PROVIDER-NAME>
        type: watsonx
        credentialsSecretRef:
          name: credentials
        url: https://ca-tor.ml.cloud.ibm.com
        projectID: <YOUR-PROJECT-ID>
        models:
          - name: ibm/granite-3-3-8b-instruct
  ols:
    defaultModel: ibm/granite-3-3-8b-instruct
    defaultProvider: <YOUR-LLM-PROVIDER-NAME>

First, log in to your OpenShift cluster before using the oc CLI. Use the following commands to complete this step.

oc login --token=<YOUR-LOGIN-TOKEN> --server=https://<YOUR-SERVER-URL>:6443

vi ./ <YOUR-LLM-PROVIDER-NAME>-llm-secret.yaml

vi ./ <YOUR-LLM-PROVIDER-NAME>-llm-config-cr.yaml

cat ./ <YOUR-LLM-PROVIDER-NAME>-llm-secret.yaml
cat ./ <YOUR-LLM-PROVIDER-NAME>-llm-config-cr.yaml

oc create/apply -f  ./ <YOUR-LLM-PROVIDER-NAME>-llm-secret.yaml
oc create/apply -f ./ <YOUR-LLM-PROVIDER-NAME>-llm-config-cr.yaml

oc get secrets/credentials -n openshift-lightspeed
oc describe secrets/credentials -n openshift-lightspeed

oc get OLSConfig/cluster -n openshift-lightspeed
oc describe OLSConfig/cluster -n openshift-lightspeed

Run the following command to check that OpenShift Lightspeed is running correctly in the cluster.

oc logs deployment/lightspeed-app-server -c lightspeed-service-api -n openshift-lightspeed | grep Uvicorn

Step 6. Access the OpenShift Lightspeed gen AI virtual assistant

In this step, you will access the gen AI virtual assistant from the OpenShift web console.

You can open the assistant in several ways:

• By clicking the floating icon in the lower-right corner of the screen.

• From the Actions dropdown after selecting a specific resource.

• From the Node Options menu in a resource list view.

The following screen shows the fully expanded virtual assistant chat window.

The following screen shows the topology view of the successfully deployed OpenShift Lightspeed virtual assistant, who is viewed from the Developer perspective.

Step 7. Ask natural language questions using the gen AI virtual assistant

In this step, you will interact with the OpenShift Lightspeed virtual assistant by asking questions in plain English. The screens show the assistant panel on the right and a terminal window on the left, which uses the oc CLI to track the complete flow from the OpenShift web console, through the Lightspeed operator, to the backend watsonx.ai model (granite-3-8b-instruct).

The following example shows a stand-alone question submitted to the assistant:

“What is an OpenShift image stream used for?”

The following screen shows the virtual assistant’s response in both the OpenShift web console and the oc CLI terminal.

Note the highlighted sections showing the assistant's reply in each window.

The following screen shows how to clear the chat history.

The following screen shows the chat being cleared, with a confirmation dialog to confirm the deletion.

The following screens show the virtual assistant's responses to a series of related questions.

Example question: How are OpenShift security context constraints used?

The following screen shows the virtual assistant’s response in both the web console and the terminal window.

The following screen shows the virtual assistant’s response to a follow-up, context-aware question:

Can I control who can use a particular SCC?

The following screen shows the virtual assistant's response to a follow-up, context-aware question.

Highlighted sections in both windows show the response details.

The following screen shows the query and response from the virtual assistant for a follow-up, context-aware question:

“Can you give me an example?”

The following screen shows a follow-up, context-aware question and the virtual assistant’s response:

“Can you give me an example?”

Review the highlighted responses in both the web console and terminal windows.

The following screen shows a question about a cluster resource, a pod, and the response from the virtual assistant.

The assistant accepts and processes the attached declarative YAML definition as an in-the-moment, in-context reference.

Pod name: lightspeed-app-server-69b799c6fb-zq2qx

Question: How many containers are running in the attached pod?

The following screen shows the virtual assistant’s response to a question about a pod resource.

The assistant uses the attached declarative YAML definition as in-the-moment, in-context input to generate its response.

The following screen shows the question related to an alert resource and the response from virtual assistant. The virtual assistant accepts in the context and in the moment resource definition as an attachment.

AlertmanagerReceiversNotConfigured What should I do about this alert?

The following screen shows the virtual assistant’s response to a question about an alert resource.

The assistant uses the attached in-the-moment, in-context resource definition to answer the question:

AlertmanagerReceiversNotConfigured – What should I do about this alert?

The following screen shows the virtual assistant’s response to a cluster alert resource. It uses the attached in-context, in-the-moment resource definition to generate the response.

The following screen shows how to start a new question after completing a chain of connected queries. This includes clearing the previous chat history and resetting the context.

Clear Chat

Start New Chat

The following screen shows the setup for a new question and the virtual assistant’s response.

The following screen shows a new question asked to the virtual assistant:

"What can Red Hat OpenShift Lightspeed do to help me?"

The assistant's response is displayed alongside.

The following screen capture shows a new question posed to the virtual assistant and the corresponding response that is displayed in the chat window.

Step 8. Review resources used in OpenShift and watsonx.ai

In this step, you will review resource usage and observability data from both the OpenShift web console and the IBM watsonx.ai Project Overview.

• As an Administrator, check API performance and kube-apiserver metrics.

• As a Developer, explore the Dashboards, Metrics, Alerts, Silences, and Events views.

• In IBM watsonx.ai, view the Project dashboard to monitor resource consumption by the backend LLM.

Acknowledgments

This tutorial was created as part of the IBM Open Innovation Community Initiative:

Better Together: Joint IBM and Red Hat assets supporting the Data and AI lifecycle in the Hybrid Cloud.

We extend our sincere thanks to our colleagues and leadership teams for their guidance, mentorship, and support throughout this journey—from early experimentation to successful validation of the concepts and technologies.

IBM Professionals:

Ahmad Azraq, Rishi Balaji, Suj Chetan Hireholi, Martin Obianwu, Nisarg Patel, Suj Perepa, Santha Kumari Ramala, Rajalaksmi S., Jay Talekar, Randall Tan, Ajay Varshney

Red Hat Professionals:

Thalia Hooker, Erik Jacobs

References and resources

• Red Hat OpenShift Lightspeed Overview

• Red Hat Announces General Availability of Red Hat OpenShift Lightspeed to Supercharge Hybrid Cloud Productivity with Generative AI

• Red Hat Solution: Getting Started with OpenShift Lightspeed