Creating an MCP (Model Context Protocol) Server on OpenShift

Large Language Models are increasingly being augmented with external tools, data sources, and actions.The Model Context Protocol (MCP) is emerging as a standard way to expose such capabilities in a structured and model-agnostic way .

In this article, we walk through how to create and run an MCP server on OpenShift, treating it like a real production microservice. And Test it using Openshift AI 3.x playground.Testing setup is mentioned in my previous article *OpenShift AI 3.0: Model Deployment, AI Asset Endpoints, and the Gen AI Playground*

Choosing the MCP Server Implementation

FastMCP is a popular framework for building servers that use the open Model Context Protocol (MCP), which provides a standardized way for AI language models (LLMs) to access external data and tools.

Local Implementation

In this example we create simple MCP server on the developer machine.

• server.py

from fastmcp import FastMCP
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
import uvicorn

mcp = FastMCP("Customer Tools")

@mcp.tool
def get_customer_details(customer_id: str) -> dict:
    return {
        "customer_id": customer_id,
        "name": "John Doe",
        "email": "john.doe@example.com",
        "status": "Gold",
        "country": "Germany",
    }

# Create BOTH ASGI apps
mcp_http_app = mcp.http_app(transport="streamable-http")
mcp_sse_app  = mcp.http_app(transport="sse")   # yes, same factory, different transport

# Pick one lifespan to pass to FastAPI.
# (Streamable-http is the critical one for OpenShift AI.)
app = FastAPI(lifespan=mcp_http_app.lifespan)

@app.get("/healthz")
def healthz():
    return {"ok": True}

# Mount endpoints
app.mount("/mcp", mcp_http_app)     # OpenShift AI uses this
app.mount("/sse", mcp_sse_app)      # MCP Inspector/debug uses this

# Added for testing with the Browser based application
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8080)

• requirements.txt

fastmcp
fastapi
uvicorn

• Run the python code.

$ python server.py 
INFO:     Started server process [4840]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8080 (Press CTRL+C to quit)

• Verify it.

# curl localhost:8080/healthz
{"ok":true}

# curl localhost:8080/sse/sse
event: endpoint
data: /sse/messages/?session_id=cb1d705707714fa79ef8ca9a4caa3b8d
: ping - 2026-01-06 08:38:59.551426+00:00

# curl -i -N -X POST localhost:8080/mcp/mcp -H "Content-Type: application/json"   -H "Accept: application/json, text/event-stream"   -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","clientInfo":{"name":"curl","version":"0.1"},"capabilities":{}}}'
HTTP/1.1 200 OK
date: Tue, 06 Jan 2026 08:37:34 GMT
server: uvicorn
cache-control: no-cache, no-transform
connection: keep-alive
content-type: text/event-stream
mcp-session-id: 96618d51b8aa4f6eb5518853046a4b87
x-accel-buffering: no
Transfer-Encoding: chunked
event: message
data: {"jsonrpc":"2.0","id":1,"result":{"protocolVersion":"2024-11-05","capabilities":{"experimental":{},"prompts":{"listChanged":true},"resources":{"subscribe":false,"listChanged":true},"tools":{"listChanged":true}},"serverInfo":{"name":"Customer Tools","version":"2.14.2"}}}

• MCP Inspector can also be run to verify it.

npx -y @modelcontextprotocol/inspector@latest
Starting MCP inspector...
⚙️ Proxy server listening on localhost:6277
🔑 Session token: 7a3326852db47d6beb07c7f1ebf9a265b02dbecacffb858375e2593224f3e0d2
   Use this token to authenticate requests or set DANGEROUSLY_OMIT_AUTH=true to disable auth

🚀 MCP Inspector is up and running at:
   http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=7a3326852db47d6beb07c7f1ebf9a265b02dbecacffb858375e2593224f3e0d2

🌐 Opening browser...

Containerizing the MCP Server

• Comment the server.py run part,And CORS settings.

from fastmcp import FastMCP
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
import uvicorn

mcp = FastMCP("Customer Tools")

@mcp.tool
def get_customer_details(customer_id: str) -> dict:
    return {
        "customer_id": customer_id,
        "name": "John Doe",
        "email": "john.doe@example.com",
        "status": "Gold",
        "country": "Germany",
    }

# Create BOTH ASGI apps
mcp_http_app = mcp.http_app(transport="streamable-http")
mcp_sse_app  = mcp.http_app(transport="sse")   # yes, same factory, different transport

# Pick one lifespan to pass to FastAPI.
# (Streamable-http is the critical one for OpenShift AI.)
app = FastAPI(lifespan=mcp_http_app.lifespan)

@app.get("/healthz")
def healthz():
    return {"ok": True}

# Mount endpoints
app.mount("/mcp", mcp_http_app)     # OpenShift AI uses this
app.mount("/sse", mcp_sse_app)      # MCP Inspector/debug uses this

# Added for testing with the Browser based application
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)

#if __name__ == "__main__":
    #uvicorn.run(app, host="0.0.0.0", port=8080)

• Create a Containerfile with ubi9/python-311 as base image.Move the execution part(uvicorn..) from server.yaml to this file.

FROM registry.access.redhat.com/ubi9/python-311

WORKDIR /opt/app-root/src

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY server.py .

EXPOSE 8080

# Run MCP SSE server and bind to 0.0.0.0:8080
# Start uvicorn directly (proxy-aware) and serve the Starlette app object "app"
CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "8080", "--proxy-headers", "--forwarded-allow-ips", "*"]

• Build it.

podman build --no-cache --platform=linux/amd64 -t customer-mcp-sse:0.1 -f Containerfile .

• Tag it and push it you available registry

podman tag localhost/customer-mcp-sse:0.1 <REPO_URL>/customer-mcp-sse:0.1
podman push <REPO_URL>/customer-mcp-sse:0.1

Kubernetes Scaffolding for Deploying an MCP Server

apiVersion: apps/v1
kind: Deployment
metadata:
  name: customer-mcp
  labels:
    app: customer-mcp
spec:
  replicas: 1
  selector:
    matchLabels:
      app: customer-mcp
  template:
    metadata:
      labels:
        app: customer-mcp
    spec:
      containers:
        - name: customer-mcp
          image: <REPO_URL>/customer-mcp-sse:0.1
          imagePullPolicy: IfNotPresent
          ports:
            - name: http
              containerPort: 8080
          readinessProbe:
            httpGet:
              path: /healthz
              port: 8080
            initialDelaySeconds: 3
            periodSeconds: 10
          livenessProbe:
            httpGet:
              path: /healthz
              port: 8080
            initialDelaySeconds: 10
            periodSeconds: 20
          resources:
            requests:
              cpu: 50m
              memory: 128Mi
            limits:
              cpu: 500m
              memory: 512Mi

apiVersion: v1
kind: Service
metadata:
  name: customer-mcp
  labels:
    app: customer-mcp
spec:
  selector:
    app: customer-mcp
  ports:
    - name: http
      port: 8080
      targetPort: 8080
  type: ClusterIP

• Apply the above definition to the Openshift cluster and create a edge route to access it externally from the cluster.

curl -k https://<Openshift_Edge_Route>/healthz
{"ok":true}

NOTE: To test the below part ,Openshift AI 3.x and LLamaStack enablement is prerequisite.

• Create a ConfigMap

apiVersion: v1
kind: ConfigMap
metadata:
  name: gen-ai-aa-mcp-servers
  namespace: redhat-ods-applications
data:
  Customer-MCP-Server: |
    {
      "url": "https://<Openshift_Edge_Route>/mcp/mcp",
      "description": "The Customer MCP server provide customer details."
    }   

• Go to OpenshiftAI Dashboard,Try it in playground.

• Request the customer details with some id .