Cloud Migration Plan

Overview

This document outlines the migration plan for the Magic e-VERSE platform from the current local server (Intel i7-9700KF, 16 GB RAM) to a cloud-hosted setup using Netcup RS 4000 G12 as the primary server, Cloudflare R2 for product image storage, and Hetzner Storage Box for off-site backups.

Tip

Estimated monthly cost: ~€34/mo — down from running a full local server with 228 GB system disk + 932 GB data disk + 1.8 TB NVMe + NAS backups.

---

Current vs New Infrastructure

Current Setup (Local Server)

Component	Spec
CPU	Intel Core i7-9700KF @ 3.60 GHz (8 cores)
RAM	16 GB DDR4 (18 GB in swap — critical)
System Disk	228 GB SSD (/dev/sda2) — 77% full
Data Disk	932 GB HDD (/dev/sdb1 → /mnt/data) — 64% full
NVMe	1.8 TB (/dev/nvme0n1p1 → /mnt/nvme) — 52% full
NAS Backup	3.5 TB + 2.7 TB (SMB mounts)
Docker Containers	53 running
Docker Images	697 GB (688 GB reclaimable)
Actual container RAM usage	~2.2 GB
Total platform data	~190 GB (excl. product images)
Product images	~220 GB (/mnt/data/pim_data/)

New Setup

Netcup RS 4000 G12

• 12 dedicated cores (AMD EPYC 9645, 2024)
• 32 GB DDR5 ECC RAM
• 1 TB NVMe SSD
• 2.5 Gbit/s network
• €27.08/mo

Cloudflare R2

• S3-compatible object storage
• Zero egress fees (free CDN delivery)
• ~250 GB for product images
• ~€3.50/mo

Hetzner Storage Box BX11

• 1 TB backup storage
• SFTP / rsync / WebDAV
• Nightly automated backups
• €3.20/mo

Total Monthly Cost

• Server: €27.08
• Images: ~€3.50
• Backups: €3.20
• Total: ~€33.78/mo

---

Why This Stack

Server: Netcup RS 4000 G12

Best price/performance ratio across all EU providers. Comparison at time of evaluation (March 2026):

Provider	Product	Cores	RAM	Storage	Network	Price/mo
Netcup	RS 4000 G12	12	32 GB DDR5	1 TB NVMe	2.5 Gbit/s	€27.08
Hetzner	CCX33 (Cloud)	8 vCPU	32 GB	240 GB	1 Gbit/s	€48.49
Hetzner	AX42 (Dedicated)	8	64 GB	1 TB	1 Gbit/s	€46.00
Contabo	VDS M	4	32 GB	240 GB	500 Mbit/s	€35.84
OVHcloud	Advance-1	6	32 GB	NVMe	1 Gbit/s	€110+

Caution

Trade-off: Netcup has no one-click scaling (unlike Hetzner Cloud) and support can be slow (24h+). But the savings of €20/mo vs Hetzner make this worthwhile for our workload. If we outgrow it, we can migrate to RS 8000 G12 (16 cores, 64 GB RAM, 2 TB NVMe, ~€58/mo).

Images: Cloudflare R2

Feature	Cloudflare R2	Firebase Storage	Hetzner Object Storage
Free storage	10 GB	5 GB	0
Egress cost	Free	€0.12/GB	€1/TB
250 GB cost	~€3.50/mo	~€8-11/mo	€6.49/mo
CDN	Global edge	Google CDN	No CDN
S3 compatible	Yes	No (proprietary SDK)	Yes

Backups: Hetzner Storage Box

• 1 TB for €3.20/mo via SFTP/rsync
• Different provider than server = true off-site backup
• 10 snapshots included
• Automated via cron + rsync

---

What Moves Where

To Netcup Server

Service	Container(s)	Current RAM
8 Commerce Tenants	backend + storefront + redis + meilisearch (×8)	~1.6 GB
Magic PIM	backend + redis	~130 MB
Spranz	backend + storefront + redis + meilisearch	~60 MB
Master Magic	backend + redis	~40 MB
PostgreSQL (shared)	magic_pim_postgres_dev	~90 MB
MySQL	mysql	~10 MB
n8n (×3)	magic_n8n, magic_n8n_2, magic_n8n_3	~120 MB
Image Services	rembg-service, preflight-service	~15 MB
Sshwifty	magic_terminal_sshwifty	~4 MB
Vaultwarden	vaultwarden	~8 MB
Magic Portal	portal (Vite + React)	~30 MB
Dashboard	dashboard API	~20 MB
DevDocs	Starlight static site	~10 MB
Total	~48 containers	~2.2 GB

To Cloudflare R2

Data	Size	Bucket
/mnt/data/pim_data/spranz/	Product images	magic-pim-images
/mnt/data/pim_data/langenberg/	Product images	magic-pim-images
/mnt/data/pim_data/svg/	SVG logos	magic-pim-images
Total	~220 GB

Stays on Local Server

Service	Reason
Omada Controller (TP-Link)	Needs LAN access for network hardware
Nextcloud + OnlyOffice	Optional — can move later or use cloud alternative

Not Migrated

Item	Reason
Old Docker images (688 GB)	Reclaimable, rebuild fresh on new server
Windows SMB shares	Local storage, not platform-related
NAS backups	Replaced by Hetzner Storage Box

---

Complete Migration Guide

Caution

Estimated downtime: 15-30 minutes (only during DNS switch in Phase 6). All preparation can be done while the current server is still running.

Phase 1: Order Infrastructure

1.1 Netcup RS 4000 G12

• Order at netcup.com/en/server/root-server/rs-4000-g12-ip-iv-12m
• Select: ip (IPv4) + iv (IPv6) + 12M (12-month contract)
• Apply voucher code at checkout: 5161nc17735292121 (1 month free)
• Wait for provisioning (usually 1-4 hours)

1.2 Cloudflare R2

• Sign up at dash.cloudflare.com
• Go to R2 → Create Bucket → Name: magic-pim-images
• Create R2 API token: R2 → Manage API Tokens → Create Token
• Note down: Account ID, Access Key ID, Secret Access Key
• Optional: set up custom domain (e.g., images.magiceverse.online)

1.3 Hetzner Storage Box

• Order BX11 (1 TB) at hetzner.com/storage/storage-box/bx11
• After provisioning, note: username (uXXXXXX), host (uXXXXXX.your-storagebox.de)
• Upload your SSH public key via Hetzner Robot panel

---

Phase 2: Prepare New Server

2.1 Initial Access & Setup

# SSH into new Netcup server (IP from Netcup SCP panel)
ssh root@<netcup-ip>

# Set hostname
hostnamectl set-hostname magic-cloud

# Update system
apt update && apt upgrade -y
apt install -y curl git htop rsync rclone unzip

# Set timezone
timedatectl set-timezone Europe/Amsterdam

2.2 Install Docker

# Install Docker Engine + Compose plugin
curl -fsSL https://get.docker.com | sh
apt install -y docker-compose-plugin

# Verify
docker --version
docker compose version

# Enable Docker on boot
systemctl enable docker

2.3 Configure Firewall

ufw default deny incoming
ufw default allow outgoing
ufw allow 22/tcp    # SSH
ufw allow 80/tcp    # HTTP
ufw allow 443/tcp   # HTTPS
ufw allow 8000/tcp  # Coolify dashboard
ufw enable

2.4 Create Directory Structure

# Match current server structure
mkdir -p /mnt/data/magic_omniverse/magic_commerce
mkdir -p /mnt/data/magic_pim
mkdir -p /mnt/data/vaultwarden
mkdir -p /opt/scripts
mkdir -p /opt/backups/databases

2.5 Configure SSH Key for Old Server Access

# Generate SSH key on new server
ssh-keygen -t ed25519 -C "magic-cloud"

# Copy to old server so we can pull data
ssh-copy-id adminmichiel1@<old-server-ip>

---

Phase 3: Upload Product Images to Cloudflare R2

Tip

Do this first — it takes the longest (~220 GB upload) and can run in the background while you prepare everything else.

3.1 Configure rclone on Current Server

# On the CURRENT (old) server
rclone config

# Enter these settings:
# name> r2
# Storage> s3
# provider> Cloudflare
# env_auth> false
# access_key_id> <your R2 access key>
# secret_access_key> <your R2 secret key>
# region> auto
# endpoint> https://<account-id>.r2.cloudflarestorage.com
# acl> (leave blank)
# Edit advanced config?> n

3.2 Sync Images

# Dry run first to verify
rclone sync /mnt/data/pim_data/ r2:magic-pim-images/ --dry-run --progress

# If looks good, do the real sync (run in tmux/screen!)
tmux new -s r2sync
rclone sync /mnt/data/pim_data/ r2:magic-pim-images/ --progress --transfers=8

# Verify file count
rclone size r2:magic-pim-images/

3.3 Set Up R2 Public Access

# In Cloudflare Dashboard:
# 1. Go to R2 → magic-pim-images → Settings
# 2. Enable "Public Access" via custom domain
# 3. Add domain: images.magiceverse.online
# 4. Cloudflare handles SSL automatically

# Test access:
curl -I https://images.magiceverse.online/spranz/5325-00.001.jpg

3.4 Update Image URL References in Code

Danger

Critical step: All image paths must be updated before switching. The current code serves images from local /mnt/data/pim_data/. After migration, images come from R2.

Files to update (in magic_development, then sync to all tenants):

File / Area	Current	New
PIM backend image serving	/mnt/data/pim_data/{supplier}/{file}	https://images.magiceverse.online/{supplier}/{file}
Storefront <Image> components	Relative paths via backend proxy	Direct R2 URL
Spranz Designer SVG paths	/mnt/data/pim_data/svg/spranz/	https://images.magiceverse.online/svg/spranz/
APLT product thumbnails	Local file read	R2 URL
Docker volume mounts	/mnt/data/pim_data:/mnt/data/pim_data:ro	Remove (no longer needed)

Tip

Approach: Create an environment variable IMAGE_BASE_URL that defaults to the R2 URL. This way you can switch back to local paths if needed. Add to each tenant’s docker-compose environment:

IMAGE_BASE_URL=https://images.magiceverse.online

---

Phase 4: Migrate Data to New Server

4.1 Database Dumps (on current server)

# PostgreSQL — all tenant databases + PIM
docker exec magic_pim_postgres_dev pg_dumpall -U postgres > /tmp/postgres_full_$(date +%Y%m%d).sql

# Check size
ls -lh /tmp/postgres_full_*.sql

# MySQL — portal, workflow data
docker exec mysql mysqldump -uroot -p<your-db-password> --all-databases > /tmp/mysql_full_$(date +%Y%m%d).sql

4.2 Transfer Code (run in tmux — takes a while)

tmux new -s migration

# Replace <netcup-ip> with actual IP throughout

# 1. Commerce tenants (the big one)
rsync -avz --progress \
  --exclude='node_modules' \
  --exclude='.next' \
  --exclude='dist' \
  /mnt/data/magic_omniverse/ root@<netcup-ip>:/mnt/data/magic_omniverse/

# 2. PIM (code only — images are in R2 now)
rsync -avz --progress \
  --exclude='node_modules' \
  --exclude='.next' \
  --exclude='uploads' \
  /mnt/data/magic_pim/ root@<netcup-ip>:/mnt/data/magic_pim/

# 3. Vaultwarden data
rsync -avz --progress /mnt/data/vaultwarden/ root@<netcup-ip>:/mnt/data/vaultwarden/

# 4. rembg + preflight services
rsync -avz --progress /home/adminwayne/rembg-service/ root@<netcup-ip>:/opt/services/rembg-service/
rsync -avz --progress /home/adminwayne/preflight-service/ root@<netcup-ip>:/opt/services/preflight-service/

# 5. Database dumps
scp /tmp/postgres_full_*.sql root@<netcup-ip>:/tmp/
scp /tmp/mysql_full_*.sql root@<netcup-ip>:/tmp/

# Note: n8n dirs (magic_n8n, magic_n8n_2, magic_n8n_3) are included
# in step 1 since they live under /mnt/data/magic_omniverse/

Caution

Exclude node_modules and build artifacts! These will be rebuilt on the new server via docker compose build. Transferring them wastes bandwidth and causes architecture mismatches.

4.3 Second Sync (just before DNS switch)

Right before you switch DNS, do a final delta sync to catch any changes:

# Quick delta sync — only transfers changed files
rsync -avz --progress --exclude='node_modules' --exclude='.next' --exclude='dist' \
  /mnt/data/magic_omniverse/ root@<netcup-ip>:/mnt/data/magic_omniverse/

# Fresh database dump
docker exec magic_pim_postgres_dev pg_dumpall -U postgres > /tmp/postgres_final_$(date +%Y%m%d_%H%M).sql
docker exec mysql mysqldump -uroot -p<your-db-password> --all-databases > /tmp/mysql_final_$(date +%Y%m%d_%H%M).sql
scp /tmp/postgres_final_*.sql /tmp/mysql_final_*.sql root@<netcup-ip>:/tmp/

---

Phase 5: Install & Configure Coolify

Tip

Why Coolify? Instead of manually managing Docker, Nginx, SSL, and deployments, Coolify gives us:

• Auto-deploy on git push to main
• Preview environments for pull requests
• Automatic SSL via built-in Traefik
• Visual dashboard for all services
• Zero-downtime deployments
• Replaces: manual Docker builds, Nginx config, Certbot, deploy scripts

See the full Monorepo + Coolify Plan for the architecture details.

All commands below run on the new Netcup server.

5.1 Install Coolify

# One-line Coolify installer
curl -fsSL https://cdn.coollabs.io/coolify/install.sh | bash

# This installs:
# - Docker (if not already installed)
# - Traefik (reverse proxy + auto-SSL)
# - Coolify dashboard
# - PostgreSQL (for Coolify's own data)

# Coolify dashboard will be available at:
# http://<netcup-ip>:8000

5.2 Initial Coolify Setup

1. Open http://<netcup-ip>:8000 in your browser
2. Create admin account
3. Go to Settings → Servers → verify the local server is detected
4. Go to Settings → DNS/SSL → set wildcard domain: *.magiceverse.online

5.3 Connect GitHub (for Phase 5B later)

Caution

Two-phase approach: We first migrate using the current multi-repo docker-compose setup (Phase 5A). GitHub connection and the monorepo are only needed in Phase 5B — you can skip this step for now and return to it later.

1. In Coolify: Sources → Add → GitHub App
2. Follow the OAuth flow to connect your GitHub account
3. Grant access to the magic-commerce repository (created in Phase 5B)

---

Phase 5A: Initial Migration (Current Setup via Coolify)

For the initial migration, we deploy using docker-compose through Coolify while keeping the current code structure.

5A.1 Set Up Shared Infrastructure in Coolify

In Coolify dashboard, create Project: “Shared Infrastructure”:

PostgreSQL:

1. Add Service → PostgreSQL 16
2. Set password: use a strong generated password
3. Expose port 5432 to the Coolify network
4. Domain: none (internal only)
5. Deploy

MySQL:

1. Add Service → MySQL 8
2. Set root password
3. Expose port 3306 to the Coolify network
4. Deploy

5A.2 Restore Databases

# Find the PostgreSQL container Coolify created
docker ps | grep postgres

# Restore PostgreSQL (use the actual container name)
docker exec -i <coolify-postgres-container> psql -U postgres < /tmp/postgres_final_*.sql

# Verify databases
docker exec <coolify-postgres-container> psql -U postgres -c "\l"

# Restore MySQL
docker exec -i <coolify-mysql-container> mysql -uroot -p<password> < /tmp/mysql_final_*.sql

Expected databases:

• magic_pim
• magic_b2b_development, magic_b2b_brinxx, magic_b2b_default
• magic_b2b_logohorloge, magic_b2b_bovisales, magic_b2b_demo
• magic_b2b_desluis, magic_b2b_jodasign

5A.3 Deploy Commerce Tenants via Coolify

For each tenant, create a Coolify Project:

Example: Project “Brinxx”

1. Add Service → Docker Compose
2. Point to the transferred code at /mnt/data/magic_omniverse/magic_commerce/magic_brinxx/
3. Set domains in Coolify:
   • Backend: admin-brinxx.magiceverse.online
   • Storefront: brinxx.magiceverse.online
4. Coolify auto-configures Traefik routing + SSL
5. Click Deploy

Repeat for all 8 tenants:

Coolify Project	Backend Domain	Storefront Domain
Development	admin-development.magiceverse.online	development.magiceverse.online
Brinxx	admin-brinxx.magiceverse.online	brinxx.magiceverse.online
Default	admin-default.magiceverse.online	default.magiceverse.online
Logohorloge	admin-logohorloge.magiceverse.online	logohorloge.magiceverse.online
Bovisales	admin-bovisales.magiceverse.online	bovisales.magiceverse.online
Demo	admin-demo.magiceverse.online	demo.magiceverse.online
De Sluis	admin-desluis.magiceverse.online	desluis.magiceverse.online
Jodasign	admin-jodasign.magiceverse.online	jodasign.magiceverse.online

5A.4 Deploy Supporting Services

PIM — Project: “Magic PIM”

1. Docker Compose from /mnt/data/magic_pim/
2. Domain: pim.magiceverse.online

Spranz — Project: “Spranz”

1. Docker Compose from /mnt/data/magic_omniverse/magic_commerce/magic_spranz/
2. Domains: backend + storefront

n8n — Project: “Automation”

1. Add 3 Docker Compose services from /mnt/data/magic_omniverse/magic_n8n/, magic_n8n_2/, magic_n8n_3/
2. Domains: n8n.magiceverse.online, n8n2.magiceverse.online, n8n3.magiceverse.online

Vaultwarden — Project: “Tools”

1. Add Service → Docker Image → vaultwarden/server:latest
2. Mount volume: /mnt/data/vaultwarden:/data/
3. Domain: vault.magiceverse.online

Sshwifty — same project “Tools”

1. Docker Compose from /mnt/data/magic_omniverse/magic_terminal/
2. Domain: terminal.magiceverse.online

rembg + preflight — same project “Tools”

1. Docker Compose from /opt/services/rembg-service/ and /opt/services/preflight-service/

Magic Portal — Project: “Portal”

1. Docker Compose from /mnt/data/magic_omniverse/magic_portal/
2. Domain: portal.magiceverse.online

Dashboard — same project “Portal”

1. Docker Compose from /mnt/data/magic_omniverse/magic_dashboard/
2. Domain: dashboard.magiceverse.online

DevDocs (Starlight) — same project “Portal”

1. Static site from /mnt/data/magic_omniverse/magic_docs/starlight/
2. Build command: npm run build → serve dist/ directory
3. Domain: devdocs.magiceverse.online

Nextcloud + OnlyOffice (optional) — Project: “Office”

1. Docker Compose (separate from docs — Nextcloud has its own compose config)
2. Domain: office.magiceverse.online

Caution

Skip Omada Controller — it manages your local network hardware and must stay on the local server.

5A.5 Verify All Services in Coolify

In the Coolify dashboard, check that all services show green/healthy status:

Shared Infrastructure
  ├── PostgreSQL 16          ✅ Healthy
  └── MySQL 8                ✅ Healthy

Brinxx
  ├── Backend                ✅ Running → admin-brinxx.magiceverse.online
  ├── Storefront             ✅ Running → brinxx.magiceverse.online
  ├── Redis                  ✅ Healthy
  └── Meilisearch            ✅ Healthy

... (repeat for all tenants)

Magic PIM
  ├── Backend                ✅ Running → pim.magiceverse.online
  └── Redis                  ✅ Healthy

Automation
  ├── n8n                    ✅ Running → n8n.magiceverse.online
  ├── n8n_2                  ✅ Running → n8n2.magiceverse.online
  └── n8n_3                  ✅ Running → n8n3.magiceverse.online

Portal
  ├── Magic Portal           ✅ Running → portal.magiceverse.online
  ├── Dashboard API          ✅ Running → dashboard.magiceverse.online
  └── DevDocs (Starlight)    ✅ Running → devdocs.magiceverse.online

Tools
  ├── Vaultwarden            ✅ Running → vault.magiceverse.online
  ├── Sshwifty               ✅ Running → terminal.magiceverse.online
  ├── rembg-service          ✅ Running
  └── preflight-service      ✅ Running

Tip

No Nginx or Certbot needed! Coolify’s built-in Traefik handles all reverse proxy routing and SSL certificate generation automatically. This replaces ~100 nginx config files and all certbot cron jobs.

---

Phase 5B: Transition to Monorepo + Auto-Deploy (After Migration)

Once Phase 5A is stable and DNS is switched, transition to the monorepo architecture for git-push deploys and preview environments.

5B.1 Create the Monorepo

# Create new GitHub repo: magic-commerce (private)
gh repo create magic-commerce --private

# Structure (see Monorepo + Coolify Plan for full details):
magic-commerce/
├── backend/           # Single Medusa codebase for all tenants
├── storefront/        # Single Next.js codebase for all tenants
├── tenants/           # Reference configs per tenant (tenant.json)
└── docker-compose.yml # Local development

5B.2 Refactor Code

1. Copy magic_development/backend/ as the single backend codebase
2. Replace hardcoded values with environment variables
3. Implement feature flag system (backend/src/features/)
4. Set up theme system for storefronts (storefront/src/themes/)
5. Push to main branch

5B.3 Reconfigure Coolify Projects

For each tenant, switch the source from local docker-compose to the GitHub monorepo:

1. In Coolify → Project: Brinxx → Backend
2. Change source: GitHub → magic-commerce repo → Build Path: /backend
3. Set environment variables:

TENANT_ID=brinxx
TENANT_TIER=enterprise
TENANT_FEATURES=technique_pricing,brand_wizard
DATABASE_URL=postgres://postgres:xxx@postgres:5432/magic_b2b_brinxx
MEDUSA_BACKEND_URL=https://admin-brinxx.magiceverse.online
IMAGE_BASE_URL=https://images.magiceverse.online

4. Enable Auto-deploy on push to main
5. Enable Preview deployments on pull requests

Repeat for all tenants — same repo, different env vars.

5B.4 Set Up Preview Environments

1. In Cloudflare DNS: add wildcard record *.preview.magiceverse.online → <netcup-ip>
2. In Coolify: enable PR previews per project
3. Flow:

Developer opens PR #42
        ↓
Coolify auto-builds preview
        ↓
Available at: pr-42.preview.magiceverse.online
        ↓
PR merged → preview auto-destroyed
PR closed → preview auto-destroyed

5B.5 Production Deploy Flow

Developer pushes to main
        ↓
Coolify detects push (GitHub webhook)
        ↓
Builds new Docker images for each tenant
        ↓
Zero-downtime container replacement
        ↓
Database migrations run automatically
        ↓
✅ Live on production domains

Tip

Rollback: Coolify keeps previous deployments. One click in the dashboard to roll back any service to a previous version.

---

Phase 6: DNS Switch (The Actual Migration Moment)

Danger

This is the point of no return. Ensure all services are running on the new server before switching DNS. You can test by adding entries to your local /etc/hosts file first.

6.1 Pre-Switch Testing via /etc/hosts

On your local machine, temporarily point domains to the new server:

# Add to /etc/hosts (your laptop/desktop)
<netcup-ip>  admin-development.magiceverse.online
<netcup-ip>  admin-brinxx.magiceverse.online
<netcup-ip>  brinxx.magiceverse.online
<netcup-ip>  pim.magiceverse.online

# Test in browser — verify everything works
# Remove /etc/hosts entries after testing

6.2 Lower DNS TTL (Do 1 Hour Before Switch)

In Cloudflare DNS dashboard:

• Set TTL of all A records to 1 minute (60 seconds)
• Wait 1 hour for old TTL to expire

6.3 Final Data Sync

# On old server — last sync before switch
docker exec magic_pim_postgres_dev pg_dumpall -U postgres > /tmp/postgres_final.sql
docker exec mysql mysqldump -uroot -p<your-db-password> --all-databases > /tmp/mysql_final.sql
scp /tmp/postgres_final.sql /tmp/mysql_final.sql root@<netcup-ip>:/tmp/

# Restore on new server (use actual Coolify container names from `docker ps`)
ssh root@<netcup-ip>
docker exec -i <coolify-postgres-container> psql -U postgres < /tmp/postgres_final.sql
docker exec -i <coolify-mysql-container> mysql -uroot -p<password> < /tmp/mysql_final.sql

6.4 Switch DNS

In Cloudflare DNS dashboard, update all A records:

Record	Old Value	New Value
*.magiceverse.online	<old-server-ip>	<netcup-ip>
portal.magiceverse.online	<old-server-ip>	<netcup-ip>
pim.magiceverse.online	<old-server-ip>	<netcup-ip>

Tip

If using Cloudflare proxy (orange cloud), DNS propagation is instant. If DNS-only (grey cloud), wait 1-5 minutes.

6.5 Verify SSL Certificates

Coolify/Traefik automatically provisions Let’s Encrypt SSL certificates once DNS points to the server. Check the Coolify dashboard to confirm all domains show a valid certificate.

6.6 Raise DNS TTL

After confirming everything works (give it 1 hour):

• Set TTL back to Auto or 3600 (1 hour)

---

Phase 7: Post-Migration Verification

7.1 Service Checklist

# Run this verification script on the new server
echo "=== Container Status ==="
docker ps --format "table {{.Names}}\t{{.Status}}" | sort

echo ""
echo "=== Memory Usage ==="
free -h

echo ""
echo "=== Disk Usage ==="
df -h /

echo ""
echo "=== Container Count ==="
echo "Running: $(docker ps -q | wc -l)"
echo "Stopped: $(docker ps -aq --filter 'status=exited' | wc -l)"

7.2 Test Every Service

Service	URL	What to Check
Development Admin	https://admin-development.magiceverse.online	Login works, products load
Brinxx Admin	https://admin-brinxx.magiceverse.online	Login, orders visible
Brinxx Storefront	https://brinxx.magiceverse.online	Products display, images load from R2
All other tenants	https://admin-{tenant}.magiceverse.online	Basic login test
PIM	https://pim.magiceverse.online	Product list loads
n8n	https://n8n.magiceverse.online	Workflows visible
Vaultwarden	https://vault.magiceverse.online	Login works
Terminal	https://terminal.magiceverse.online	SSH connection works
Portal	https://portal.magiceverse.online	Dashboard loads, tenant list visible
Dashboard	https://dashboard.magiceverse.online	API responds, dev projects accessible
DevDocs	https://devdocs.magiceverse.online	Pages load
Product images	https://images.magiceverse.online/spranz/5325-00.001.jpg	Image loads

7.3 Monitor for 24 Hours

# Install monitoring
apt install -y glances

# Watch live
glances

# Check Docker logs for errors
docker ps -q | xargs -I {} docker logs {} --tail=5 2>&1 | grep -i "error\|fatal\|crash"

---

Phase 8: Cleanup

8.1 Old Server

• Keep running for 1 week as fallback
• After 1 week with no issues, stop all Docker containers
• Keep database dumps as archive
• Repurpose server for local dev / Omada Controller only

8.2 New Server Optimization

# Clean up Docker build cache
docker builder prune -f

# Remove dangling images
docker image prune -f

# Check final disk usage
df -h /

8.3 Update SSH Config

On your local machine (~/.ssh/config):

Host magic-cloud
    HostName <netcup-ip>
    User root
    IdentityFile ~/.ssh/id_ed25519
    StrictHostKeyChecking no

8.4 Update CLAUDE.md

Update the server documentation to reflect the new infrastructure — IP addresses, paths, and connection details.

---

Backup Strategy

Local Backups (on Netcup server)

# Daily database backup script — /opt/scripts/backup-databases.sh
#!/bin/bash
BACKUP_DIR="/opt/backups/databases"
DATE=$(date +%Y%m%d_%H%M)
mkdir -p $BACKUP_DIR

# PostgreSQL
docker exec magic_pim_postgres_dev pg_dumpall -U postgres | gzip > $BACKUP_DIR/postgres_$DATE.sql.gz

# MySQL
docker exec mysql mysqldump -uroot -p<your-db-password> --all-databases | gzip > $BACKUP_DIR/mysql_$DATE.sql.gz

# Keep last 7 days
find $BACKUP_DIR -name "*.sql.gz" -mtime +7 -delete

echo "Backup completed: $DATE"

Remote Backups (to Hetzner Storage Box)

# Rsync to Hetzner Storage Box — /opt/scripts/backup-remote.sh
#!/bin/bash
STORAGEBOX="uXXXXXX@uXXXXXX.your-storagebox.de"

# Sync database backups
rsync -avz --delete /opt/backups/databases/ $STORAGEBOX:backups/databases/

# Sync docker-compose files and configs
rsync -avz --delete /mnt/data/magic_omniverse/magic_commerce/*/docker-compose*.yml $STORAGEBOX:backups/compose/
rsync -avz --delete /opt/coolify/ $STORAGEBOX:backups/coolify/

echo "Remote backup synced: $(date)"

Cron Schedule

# /etc/crontab additions
# Local DB backup — every 6 hours
0 */6 * * * root /opt/scripts/backup-databases.sh >> /var/log/backup.log 2>&1

# Remote sync — daily at 3 AM
0 3 * * * root /opt/scripts/backup-remote.sh >> /var/log/backup-remote.log 2>&1

R2 Image Backup

Product images in Cloudflare R2 are the source of truth. R2 has built-in durability (99.999999999%). For extra safety:

# Weekly sync of R2 to Hetzner Storage Box
0 4 * * 0 root rclone sync r2:magic-pim-images/ /tmp/r2-mirror/ && rsync -avz --delete /tmp/r2-mirror/ $STORAGEBOX:backups/images/ && rm -rf /tmp/r2-mirror/

---

Scaling Path

When	Action	New Specs	Price
Current	RS 4000 G12	12 cores, 32 GB, 1 TB	€27/mo
Need more disk	Add Netcup Block Storage	+500 GB NVMe	+€5.95/mo
Need more RAM/CPU	Migrate to RS 8000 G12	16 cores, 64 GB, 2 TB	€58/mo
High traffic	Add Cloudflare CDN (free)	Cache storefronts at edge	€0
Need HA/scaling	Migrate to Hetzner Cloud	CCX series + load balancer	Variable

---

Important Notes

Danger

Image URL Migration: Before switching, all product image references in the database and codebase must be updated to use the R2 custom domain (e.g., https://images.magiceverse.online/spranz/5325-00.001.jpg). This affects:

• PIM backend image serving
• Storefront <img> tags and Next.js Image components
• Spranz Designer SVG paths
• APLT product image references

Caution

Beta Brinxx Server (beta.brinxx.nl, 37.97.132.23) is a separate Hetzner VPS and is NOT part of this migration. It continues to run independently with its own 96 GB disk.

Tip

Voucher codes for Netcup RS 4000 G12 (1 month free): 5161nc17735292121 or 5161nc17724492130. Apply during checkout.