Opa Jan's Homelab

restore-from-scratch

3 min read

How to Restore from Scratch

Complete disaster recovery - rebuild homelab from nothing.

When to use

  • Complete Proxmox host failure
  • Migrating to new hardware
  • Catastrophic data loss

Prerequisites

  • Proxmox VE installed on new hardware
  • Network configured (same subnet: 192.168.144.0/23)
  • Backup of /lxcdata/ (service data)
  • Backup of LXC/VM configs
  • Copy of ~/dev/homelab-docker/ repository

Steps

1. Prepare Proxmox host

Install Proxmox VE and configure:

  • Network: 192.168.144.0/23, gateway .1
  • Storage: ZFS for /lxcdata and VM storage
  • Network bridge: vmbr0

2. Create base template (CT 902)

# Create Arch Linux LXC
# (Download Arch Linux template in Proxmox GUI first)
pct create 902 local:vztmpl/archlinux-lxc-template_*.tar.zst \
  --hostname lxc-base \
  --cores 2 \
  --memory 2048 \
  --net0 name=eth0,bridge=vmbr0,ip=dhcp \
  --features nesting=1,keyctl=1
 
# Start template
pct start 902
 
# Disable sandbox in pacman.conf
pct exec 902 -- sed -i '/\[options\]/a DisableSandbox' /etc/pacman.conf
 
# Install Docker
pct exec 902 -- pacman -Sy docker docker-compose
pct exec 902 -- systemctl enable docker
 
# Stop template
pct stop 902

3. Restore service data

# Copy service data from backup
rsync -av /backup/homelab/ /lxcdata/

4. Recreate containers

For each service:

# Clone template
pct clone 902 <CT_ID> --hostname <service> --full
 
# Set resources
pct set <CT_ID> --cores <cores> --memory <ram>
 
# Configure network
pct set <CT_ID> -net0 name=eth0,bridge=vmbr0,firewall=1,gw=192.168.144.1,ip=<IP>/23
 
# Mount data directory
pct set <CT_ID> -mp0 /lxcdata/<service>,mp=/data
 
# Enable features
pct set <CT_ID> -features nesting=1,keyctl=1
 
# Auto-start
pct set <CT_ID> -onboot 1
 
# Add AppArmor workaround
cat >> /etc/pve/lxc/<CT_ID>.conf << 'EOF'
lxc.apparmor.profile: unconfined
lxc.mount.entry: /dev/null sys/module/apparmor/parameters/enabled none bind 0 0
EOF
 
# Start and deploy
pct start <CT_ID>
pct exec <CT_ID> -- systemctl start docker
pct exec <CT_ID> -- bash -c 'git clone https://github.com/opajanvv/homelab-docker.git /opt/homelab-docker'
pct exec <CT_ID> -- bash -c 'cd /opt/homelab-docker/<service> && chmod +x install.sh && ./install.sh'

5. Recreate VMs (Immich, Home Assistant)

For each VM:

  • Create VM with appropriate resources
  • Restore VM disk from backup or reinstall OS
  • Configure networking

6. Restore databases

MariaDB (WordPress):

# After wordpress-db container is running
cat mariadb-backup.sql | pct exec 128 -- bash -c 'docker exec -i mariadb mysql -u root -p'

PostgreSQL (Planka):

cat planka-backup.sql | pct exec 122 -- bash -c 'docker exec -i postgres psql -U planka'

7. Recreate routing

Cloudflare Tunnel:

  • Log into Cloudflare Zero Trust dashboard
  • Recreate tunnel or reconfigure existing tunnel
  • Add all routes (see network documentation)

Lanproxy (Caddy):

  • Reinstall lanproxy container
  • Restore Caddyfile from backup
  • Reload configuration

8. Verify all services

Go through each service:

  • Container running
  • Service responding
  • External access working
  • Internal access working

Service Priority Order

Restore in this order:

  1. Infrastructure: wordpress-db, lanproxy, cloudflared
  2. Critical services: n8n, planka
  3. Web services: grav, wordpress sites
  4. Other services: jellyfin, ai, syncthing, immich, home-assistant

Estimated Times

  • Proxmox setup: 1-2 hours
  • Template creation: 30 minutes
  • Per service: 10-15 minutes
  • Total: 4-6 hours for full restore

Verification

  • All containers running: pct list shows “running”
  • All services accessible via web
  • External routes working
  • Data intact (check a few key services)
  • Backups working

Backlinks