flowmaster-server

name	flowmaster-server
description	FlowMaster server configuration and infrastructure
disable-model-invocation	false

FlowMaster Server Configuration Skill

Server Architecture

Servers Overview

Server	Role	Deployment	Status
Dev-01	Primary K3S cluster	K3S	UP
Dev-02	High-perf dev, agent pods	K3S	UP
Production	app.flow-master.ai	Docker Compose	UP

SSH Access: dev-01 (user: ben) or dev-01-root (user: root) OS: Ubuntu 24.04.3 LTS

---

Dev-01 Server — PRIMARY

3 FlowMaster Environments on K3S

Production Environment (flowmaster namespace)

• Access: Direct kubectl only (nginx may point elsewhere)
• Status: Fully operational, separate from default nginx routing

Development Environment (flowmaster-dev namespace) — DEFAULT VIA NGINX

• Access: Public URL via server IP
• Status: Login working, dev credentials active

Test Environment (flowmaster-test namespace)

• Access: Port 8080 (may be blocked by firewall)
• Status: Isolated test environment with separate databases

K3S Shared Resources

• Local Registry: localhost:30500 (K3S NodePort)

Deployed Services (29):

Service	Port	Status
agent-service	9016	Running
ai-agent	9006	Running
api-gateway	9000	Running
auth-service	8002	Running
bac-marketplace	9015	Running
business-rules-engine	8018	Running
document-intelligence	9002	Running
dxg-service	9011	Running
engage-app	3001	Running
event-bus	9013	Running
execution-engine	9005	Running
external-integration	9014	Running
frontend	3000	Running
human-task	9006	Running
knowledge-hub	8009	Running
legal-entity-service	8014	Running
manager-app	3001	Running
mcp-server	9000	Running
notification	9009	Running
process-analytics	9014	Running
process-design	9003	Running
process-designer	3002	Running
process-linking	8021	Running
process-versioning	8020	Running
process-views	8019	Running
prompt-engineering	8012	Running
scheduling	9008	Running
service-registry	8001	Running
websocket-gateway	9010	Running

ExternalName services: arangodb, postgres, redis (pointing to databases-test namespace)

SDX Platform (Part of FlowMaster, currently standalone Docker)

• Location: /opt/sdx/
• Containers: 3 (backend, frontend, mcp-server)
• Note: Should be migrated into K3S cluster

Shared Infrastructure

• ArangoDB: Shared by FlowMaster + SDX
• Nginx: Reverse proxy on port 80 to K3S ClusterIPs

---

Plane CE (Separate — Project Management Tool, NOT part of product)

• Location: /opt/plane/
• Containers: 13 (proxy, web, admin, space, live, api, worker, beat-worker, db, redis, mq, minio, mcp-server)
• Workspace: flowmaster

---

Network Architecture

Routing Pattern:

Internet → Nginx (port 80) → target namespace
  ├── / → frontend ClusterIP (look up live)
  ├── /api/ → api-gateway ClusterIP (look up live)
  └── /ws/ → websocket-gateway ClusterIP (look up live)

WARNING: ClusterIPs are hardcoded in /etc/nginx/sites-enabled/flowmaster. They will change if K3S services are recreated.

Nginx Config Location

• /etc/nginx/sites-enabled/flowmaster
• Proxy timeouts: 3600s for API (AI operations), 86400s for WebSocket

---

Deployment Method (CURRENT)

How Services Were Deployed

Code was deployed via manual docker pipeline, NOT via GitLab CI/CD:

1. Code written locally (macOS arm64)
2. Docker images built with --platform linux/amd64
3. Images transferred via docker save | gzip | scp
4. Loaded on server via docker load
5. Pushed to K3S local registry (localhost:30500)
6. K3S deployments updated via kubectl set image

Code has NOT been pushed to GitLab repositories.

CI/CD Pipeline (Available but not used for current deployment)

Platform: GitLab CI/CD

Pipeline Stages:

detect-changes -> test -> build -> deploy

Required GitLab CI/CD Variables:

• SSH_PRIVATE_KEY — SSH key for server access (type: file)
• DEMO_SERVER_1_HOST — Server IP
• Database vars: ARANGO_HOST, POSTGRES_GLOBAL_HOST, REDIS_HOST + credentials
• Security vars: JWT_SECRET, ENCRYPTION_KEY, SECRET_KEY
• API keys: GEMINI_API_KEY, OPENAI_API_KEY, OPENROUTER_API_KEY

All K3S deployments have imagePullPolicy: Always.

---

Directory Structure (Dev-01 Server)

/opt/flowmaster-deployments/    # Legacy staging compose (not actively used)
    docker-compose.staging.yml

/opt/sdx/                       # SDX deployment
    docker-compose.yml

/opt/plane/                     # Plane CE
    docker-compose.yaml

/opt/plane-mcp-server/          # Plane MCP Server
    docker-compose.yml

K3S manages FlowMaster services (not Docker Compose).

---

Production Server

• URL: app.flow-master.ai
• Status: UP (HTTP 200)
• Deployment: Docker Compose, production branch
• SSL: Let's Encrypt

---

Common Troubleshooting Commands

K3S

# Check all pods
kubectl get pods -n flowmaster

# Check specific pod logs
kubectl logs -n flowmaster deployment/api-gateway --tail 100

# Restart a deployment
kubectl rollout restart deployment/api-gateway -n flowmaster

# Update image
kubectl set image deployment/api-gateway api-gateway=localhost:30500/flowmaster/api-gateway:<tag> -n flowmaster

# Check services and ClusterIPs
kubectl get svc -n flowmaster

# Check resource usage
kubectl top pods -n flowmaster

Nginx

# Test config
sudo nginx -t

# Reload
sudo systemctl reload nginx

# View config
cat /etc/nginx/sites-enabled/flowmaster

Docker (Plane/SDX)

# Check Plane containers
docker ps | grep plane

# Check SDX containers
docker ps | grep sdx

# Check ArangoDB
docker ps | grep arango

---

When to Use This Skill

1. Deploying FlowMaster — understanding K3S cluster, image pipeline
2. Managing infrastructure — nginx, SSL, port allocation
3. Debugging container issues — pod logs, restarts, ClusterIP changes
4. CI/CD pipeline work — GitLab runner, deployment jobs
5. Server access — SSH patterns, directory structure

For live server IPs, service status, and image tags, use commander-mcp: get_context_servers, get_context_services