CTERA Edge Filer MCP Server for Edge Caching

Overview

The CTERA Edge Filer MCP server implements a lightweight Model Context Protocol (MCP) gateway that delivers intelligent edge caching and multiprotocol file access for hybrid deployments. It sits at remote or core sites to provide low-latency access to shared file data while enforcing secure, authenticated access and minimizing bandwidth between edge sites and central storage.

Developers use the MCP server to integrate application models, orchestration systems, or AI agents with edge-resident file resources. The server exposes a simple API and can proxy or cache files for common protocols (SMB/NFS/REST) so clients benefit from local performance and consistent global data access patterns.

Features

• Edge caching for frequently accessed files to reduce WAN traffic
• Multiprotocol access: supports file access via RESTful API, and can be configured to front SMB/NFS exports
• Secure transport and authentication (TLS, token-based auth, pluggable identity backends)
• Configurable cache policies: TTLs, prefetch, size limits, eviction strategies
• Lightweight deployment: containerized image and small memory footprint for branch/edge hardware
• Health and metrics endpoints for observability (Prometheus-compatible)
• Admin API for cache control, invalidation, and status inspection

Installation / Configuration

This section shows common ways to deploy the MCP server. Replace example values with your environment-specific settings.

1. Clone the repository:

git clone https://github.com/ctera/mcp-ctera-edge.git
cd mcp-ctera-edge

2. Build and run with Docker:

# build image
docker build -t ctera/mcp-edge:latest .

# run container with minimal config
docker run -d \
  --name mcp-edge \
  -p 8443:8443 \
  -e MCP_SERVER_HOST=0.0.0.0 \
  -e MCP_CACHE_DIR=/var/lib/mcp/cache \
  -e MCP_TLS_CERT=/certs/tls.crt \
  -e MCP_TLS_KEY=/certs/tls.key \
  -v /srv/mcp/cache:/var/lib/mcp/cache \
  -v /etc/ssl/certs:/certs:ro \
  ctera/mcp-edge:latest

3. Example Kubernetes Deployment (snip):

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mcp-edge
spec:
  replicas: 1
  selector:
    matchLabels:
      app: mcp-edge
  template:
    metadata:
      labels:
        app: mcp-edge
    spec:
      containers:
      - name: mcp-edge
        image: ctera/mcp-edge:latest
        ports:
        - containerPort: 8443
        env:
        - name: MCP_CACHE_DIR
          value: "/var/lib/mcp/cache"
        volumeMounts:
        - name: cache
          mountPath: /var/lib/mcp/cache
      volumes:
      - name: cache
        persistentVolumeClaim:
          claimName: mcp-cache-pvc

4. Minimal configuration example (config.yaml):

server:
  host: 0.0.0.0
  port: 8443
  tls:
    cert_file: /certs/tls.crt
    key_file: /certs/tls.key

cache:
  dir: /var/lib/mcp/cache
  max_size_gb: 50
  eviction_policy: lru
  prefetch: true

auth:
  mode: token
  token_secret: "REPLACE_WITH_SECURE_SECRET"

Configuration options (summary):

Available Tools

• mcpctl — A small CLI shipped with the repo to query status, invalidate cache keys and force prefetch.
• REST API — Exposes endpoints for file retrieval, cache control and metrics (see /health, /metrics, /cache/, /files/).
• Prometheus metrics — /metrics endpoint for scraping cache hit/miss counts and latency histograms.
• Example integrations — Example scripts in the repo demonstrating fetching files, prefetch scheduling and token generation.

Example CLI usage:

# query health
./mcpctl --endpoint https://edge.example.local:8443 health

# list cache status
./mcpctl --endpoint https://edge.example.local:8443 cache list

# invalidate a file
./mcpctl --endpoint https://edge.example.local:8443 cache invalidate --path /shared/data/report.pdf

Use Cases

• Remote office file acceleration: Branch offices get local, fast access to shared repositories (home directories, shared data) while the MCP server synchronizes and caches hot content from the core.
• CI/CD artifact caching: Developer pipelines at remote sites can pull large dependencies and build artifacts from a nearby cache to reduce build time and WAN usage.
• VDI/user profile optimization: Host user profiles and session data locally to speed login times for virtual desktops while keeping centralized control.
• Disaster recovery seeding and bandwidth smoothing: Prefetch critical datasets to edge sites during off-peak windows to ensure availability during disruptions.
• AI/ML model context access: Provide low-latency read access to model inputs (data files) for inference services running at the edge.

Getting Help / Next Steps

• Read the included API docs in the repository for endpoint details and example requests.
• Use the provided sample configurations to adapt cache sizing and eviction to your workload.
• Monitor cache hits and WAN egress via the /metrics endpoint to tune prefetch and TTL settings.