> Software > Manage > Integration > MCP Servers > Kubernetes MCP server

Deploy the Kubernetes MCP server and connect an AI client

Deploy the Kubernetes MCP server and connect an AI client to manage Commvault components in your Kubernetes environment.

Requirements

  • A Kubernetes cluster with sufficient resources

  • kubectl configured with access to your Kubernetes cluster

  • Helm installed on the system where you will run the setup script

  • Docker or Node.js, depending on deployment mode

Deploy the MCP server

Download the MCP server and run the setup wizard.

  1. Clone the repository:

    git clone https://github.com/Commvault/commvault-k8s-devops-mcp.git

  2. Go to the repository directory:

    cd commvault-k8s-devops-mcp

  3. Run the setup script:

    • For Linux or macOS: chmod +x setup.sh && ./setup.sh

    • For Windows: .\setup.ps1

  4. Select a deployment mode based on your use case.

    Run the MCP server on a single system using stdio transport. Use this mode for testing or evaluation.

    The setup script configures the MCP server locally and generates a configuration file (mcp.json) for your AI client.

    Deploy the MCP server in a Kubernetes cluster using HTTP or SSE transport. Use this mode for shared or production environments.

    During setup, provide the following information:

    • Container registry, image name, and tag for the MCP server

    • Kubernetes namespaces for the MCP server and Commvault workloads

    • Image registry for Commvault components

    • External hostname and port (default: 8403)

    • Authentication method: use static-bearer

    For configuration details, see Configuration Reference section of the Kubernetes MCP Server page on GitHub.

  5. After setup completes, use the generated configuration file to connect your AI client.

    • .vscode/mcp.json, for GitHub Copilot (Visual Studio Code)

    • claude_desktop_config.json for Claude Desktop.

Connect an AI client

Configure your AI client using the generated configuration file.

  1. Copy the generated mcp.json file to your workspace.

  2. Open the workspace in Visual Studio Code.

  3. In GitHub Copilot, verify that the commvault-k8s server appears under MCP servers.

  4. Start using Copilot to manage your Kubernetes environment.

  1. Go to Settings > Developer > Edit Config.

  2. Copy the contents of claude_desktop_config.json into your configuration.

  3. Restart Claude Desktop.

  4. Start a conversation to manage your Kubernetes environment.

Configure the MCP server in your client settings:

  • Type: sse for Kubernetes deployment or stdio for local deployment

  • URL: provided in the setup output. For example, https://mcp.company.com/sse.

  • Authorization: Bearer token provided during setup.

After you connect your AI client, you can start deploying, monitoring, and troubleshooting Commvault components in your Kubernetes environment. For more information about workflows and tools, see the Tool Reference section of the Kubernetes MCP Server page on GitHub.

The MCP server also includes advanced namespace management features for multi-user and multi-tenant deployments. For more information, see the Namespace Management section of the Kubernetes MCP Server page on GitHub.

×

Loading...