Skip to main content
The bootstrap commands set up your local Kubernetes environment with all required infrastructure components. Choose between fast development mode or full production-parity mode.

​
bootstrap

Dev-fast bootstrap with kindnetd networking (no Cilium). Optimized for quick local development with warm cluster support.

​
Syntax

bootstrap [--clean] [--full]

​
Options

​
--clean
flag
Force full rebuild by destroying existing cluster and clearing state cache
​
--full
flag
Delegate to bootstrap-full for Cilium-based networking (production parity mode)

​
Features

  • Warm cluster support: Automatically detects cluster state and only rebuilds what changed
  • Hash-based caching: Tracks cluster config and manifest changes to minimize unnecessary rebuilds
  • Single control-plane node: Minimal resource footprint
  • kindnetd networking: Fast startup without Cilium overhead
  • Parallel execution: 4-phase pipeline with concurrent operations

​
Workflow

The bootstrap process runs in 4 phases:
  1. Phase 1: Preparation (parallel)
    • Create kind cluster
    • Generate nixidy manifests
    • Fetch/build OTel Collector image
    • Pull container images
  2. Phase 2: Image Load
    • Load images into kind cluster
    • Load custom OTel Collector
  3. Phase 3: Deploy Services (parallel)
    • Deploy PostgreSQL (early start)
    • Deploy Garage (S3-compatible storage)
    • Deploy observability stack (Prometheus, Grafana, Loki, Tempo)
    • Deploy Traefik ingress
    • Deploy Cloudflared tunnel (if configured)
  4. Phase 4: Wait for Readiness
    • Wait for all pods to become ready (parallel checks)

​
Examples

# Standard bootstrap (warm cluster aware)
bootstrap

# Force clean rebuild
bootstrap --clean

# Use full Cilium mode instead
bootstrap --full

​
Warm Cluster Behavior

The command intelligently handles existing clusters:
  • Cluster config unchanged + manifests unchanged: Quick health check only
  • Cluster config unchanged + manifests changed: Reapply manifests without cluster rebuild
  • Cluster config changed: Full rebuild with new cluster
  • Cluster unhealthy: Full rebuild

​
Output

After successful bootstrap: 
=== Bootstrap complete (dev-fast) ===
Mode: kindnetd (no Cilium)
Node: single control-plane

Next: cd microservice-app && tilt up

  Grafana:      http://localhost:30300  (admin/admin)
  Prometheus:   http://localhost:30090
  Alertmanager: http://localhost:30093
  Traefik:      http://localhost:30081

​
bootstrap-full

Full production-parity bootstrap with Cilium CNI and Istio service mesh. Includes ArgoCD for GitOps and 2 worker nodes.

​
Syntax

bootstrap-full [--clean]

​
Options

​
--clean
flag
Force full rebuild by destroying existing cluster and clearing state cache

​
Features

  • Cilium CNI: eBPF-based networking with Hubble observability
  • Istio ambient mode: Service mesh with L4 and L7 capabilities
  • ArgoCD: GitOps continuous delivery
  • Gateway API: Kubernetes Gateway API v1.5.0 CRDs
  • Multi-worker setup: 2 worker nodes for realistic testing
  • Warm cluster support: Same intelligent caching as bootstrap

​
Workflow

Full bootstrap runs in 4 phases:
  1. Phase 1: Preparation (parallel)
    • Create kind cluster (3 nodes: 1 control-plane, 2 workers)
    • Generate nixidy manifests
    • Build OTel Collector image
    • Pull container images (Cilium, Istio, application images)
  2. Phase 2: Network Setup (sequential)
    • Load Cilium and OTel images into kind
    • Install Cilium with Hubble UI
    • Install Istio ambient mode with tracing
    • Apply Gateway API CRDs
    • Start PostgreSQL early (background)
    • Load remaining images in background
  3. Phase 3: Deploy Services (parallel)
    • Deploy ArgoCD
    • Deploy Garage (S3-compatible storage)
    • Deploy observability stack
    • Deploy Cloudflared tunnel (if configured)
  4. Phase 4: Wait for Readiness (parallel)
    • Wait for ArgoCD server
    • Wait for PostgreSQL
    • Wait for Grafana
    • Wait for Prometheus

​
Examples

# Standard full bootstrap
bootstrap-full

# Force clean rebuild
bootstrap-full --clean

​
Output

After successful bootstrap: 
=== Bootstrap complete (full) ===
ArgoCD will sync the remaining applications automatically.

  ArgoCD:       http://localhost:30080  (admin/<password>)
  Grafana:      http://localhost:30300  (admin/admin)
  Prometheus:   http://localhost:30090
  Alertmanager: http://localhost:30093
  Hubble UI:    http://localhost:31235
  Traefik:      http://localhost:30081

​
Cilium Configuration

Cilium is installed with Istio coexistence settings: 
cni.exclusive: false              # Chain with Istio CNI
socketLB.hostNamespaceOnly: true  # Avoid Istio traffic conflicts
kubeProxyReplacement: false       # Safe Istio coexistence
hubble.enabled: true              # Network observability
hubble.ui.nodePort: 31235         # Hubble UI access

​
Istio Configuration

Istio is installed in ambient mode with:
  • Distributed tracing enabled (OpenTelemetry)
  • OTel Collector endpoint: otel-collector.observability.svc.cluster.local:4317
  • Waypoint proxy in microservices namespace
  • Ambient dataplane mode for labeled namespaces

​
full-bootstrap

Legacy command. Use bootstrap-full instead for current full setup.
Lightweight bootstrap without Istio and ArgoCD. Single worker node for reduced memory usage.

​
Syntax

full-bootstrap

​
Features

  • Cilium CNI with Hubble
  • Single worker node (reduced from 2)
  • No Istio service mesh
  • No ArgoCD (use Tilt for app deployment)
  • Memory-optimized configuration

​
Use Case

Ideal for development environments with memory constraints where service mesh and GitOps are not required.

Build docs developers (and LLMs) love

Get started for freeTalk to us