Kubernetes

Deploy your agent to a Kubernetes cluster.

Before you begin

1. Follow the Get started guide to set up agentregistry and start the agentregistry daemon.

2. Publish an MCP server.

3. Create a Kubernetes cluster. For example, you can use the following command to create a kind cluster.

   kind create cluster --name agentregistry

4. Make sure that your current kubeconfig context points to the cluster that you want to use.

   kubectl config get-contexts

5. Follow the Quickstart in the kagent OSS documentation. Agentregistry uses kagent for bootstrapping during an MCP server deployment.

6. Local setups only: If you built the docker images locally without pushing them to a registry, load the MCP server image to your kind or minikube cluster. The following command assumes that you use kind and that your cluster is named agentregistry.

   kind load docker-image my-mcp-server:latest --name agentregistry 

Deploy the MCP server

1. Deploy the MCP server to your cluster.

   arctl deployments create user/my-mcp-server \
    --type mcp \
    --provider-id kubernetes-default \
    --namespace default \
    --version 0.1.0

   Example output:

   Deploying server...
   
   Deployed user/my-mcp-server (v0.1.0) with providerId=local
   
   Server deployment recorded. The registry will reconcile containers automatically.
   Agent Gateway endpoint: http://localhost:21212/mcp
   
   If the deployment fails, a deployment entry is still created in the database and set to failed. You can view failed deployments by using the arctl deployments list command. Note that you cannot re-deploy a failed deployment to fix it. Instead, remove the failed deployment with arctl deployments delete <deployment-ID> and then re-run the arctl deployments create command.

2. Verify that the MCP server is up and running.

   kubectl get pods | grep mcp    

   Example output:

   NAME                                           READY   STATUS    RESTARTS   AGE
   user-my-mcp-server-1a1a3db9-9b5fbbd94-5c6k9    1/1     Running              61s
   

3. Optional: Review the MCPServer resource that was created in your cluster.

   kubectl get mcpserver -o yaml

Cleanup

You can remove a deployment from the UI or CLI.