Skip to main content

🏗️ Platform Engineering with TunaOS

TunaOS is built for platform engineers. Every aspect — from the immutable bootc foundation to the tooling ecosystem — is designed to make infrastructure reliable, reproducible, and manageable at scale.

Why TunaOS for Platform Engineering​

NeedTunaOS Solution
Immutable OSBootc images — atomic updates and rollbacks
Reproducible buildsContainer-native images built from Containerfiles
Fleet managementbootc switch / bootc upgrade across machines
CI/CDGitHub Actions build pipeline with OCI registry
Developer toolingHomebrew pre-installed, Tavern GUI, Bluefin CLI
VM managementCorral — QEMU/KVM and KubeVirt backends
USB provisioningTacklebox — multi-boot media creation
GPU workloads-nvidia variants with NVIDIA/CUDA (formerly GDX)
Desktop choiceGNOME, KDE, COSMIC, Niri, XFCE

Immutable Infrastructure​

TunaOS uses bootable containers (bootc) — the same image-based approach as OpenShift CoreOS:

# Every machine runs from an OCI image
sudo bootc status

# Updates are image pulls + reboots
sudo bootc upgrade && sudo systemctl reboot

# Rollback is one command
sudo bootc rollback && sudo systemctl reboot

This means your entire fleet is defined by a container image digest. No configuration drift, no snowflake servers.

CI/CD Pipeline​

TunaOS images are built in GitHub Actions and published to ghcr.io. The pipeline:

  1. PR → build test image → run smoke tests
  2. Merge to main → build all variants → push to registry
  3. Tagged release → build + sign + push + create GitHub release

You can fork this pipeline for your own custom images. See Building TunaOS and CI/CD for details.

Fleet Management​

Manage multiple machines with bootc:

# SSH into each machine and update
for host in dev-box-1 prod-box-1 prod-box-2; do
  ssh "$host" "sudo bootc upgrade && sudo systemctl reboot"
done

# Pin all machines to the same digest
for host in prod-box-1 prod-box-2; do
  ssh "$host" "sudo bootc switch ghcr.io/tuna-os/yellowfin:gnome@sha256:abc123..."
  ssh "$host" "sudo systemctl reboot"
done

Developer Workstations​

Provision consistent developer environments:

# Each developer switches to the team's custom image
sudo bootc switch ghcr.io/my-org/dev-image:latest

# Pre-installed tooling via Brewfile
brew bundle install --file team.Brewfile

# Local VMs with Corral
corral create dev-env --qemu --container-disk quay.io/containerdisks/fedora:42
corral ssh dev-env

Provisioning USB Media with Tacklebox​

Create multi-boot USB drives for on-prem provisioning:

# Recipe with multiple TunaOS variants + diagnostic tools
tacklebox build recipe.json --img provisioning.img
sudo dd if=provisioning.img of=/dev/sdX bs=4M status=progress

See the Tacklebox Getting Started guide.

VM Infrastructure with Corral​

Manage VMs across laptop and cluster with one tool:

# Local dev VM
corral create dev --qemu --container-disk quay.io/containerdisks/fedora:42

# Production VM on Kubernetes
corral create web --kubevirt --container-disk quay.io/containerdisks/fedora:42

# Same commands, same workflow
corral ssh dev
corral ssh web

See the Corral Getting Started guide.

Security​

  • Atomic updates — no partial update states
  • Image signing — all TunaOS images are signed with cosign
  • SBOM — software bill of materials generated per image
  • Immutable /usr — system files cannot be modified at runtime
  • User data in /var and /etc — clean separation of OS and data

Monitoring​

# Check deployment status across machines
bootc status --json

# Monitor update availability
systemctl status bootc-fetch-apply-updates.timer

# View boot logs
journalctl -b -o short-monotonic

See Also​