Prerequisites

What You’re Installing

Component	Description
API Service	Long-running Kubernetes Deployment
Executor Pool	Autoscaling workers for workflow execution
CockroachDB	Internal distributed database (optional, can use external)
Redis	Job queue and cache

Required External Service

• S3-compatible storage — AWS S3, Cloudflare R2, Google Cloud Storage, MinIO, etc.

Required Tools

For Local Development (k3d)

# macOS
brew install k3d kubectl helm docker

# Linux
curl -s https://raw.githubusercontent.com/k3d-io/k3d/main/install.sh | bash
# Install kubectl and helm separately

For Production

• kubectl — Kubernetes CLI
• helm — Package manager for Kubernetes

Cluster Requirements

Local Development

k3d creates everything you need automatically:

cd apps/backend/kubernetes
./scripts/k3d-setup.sh

Production Cluster

• Any Kubernetes distribution (EKS, GKE, AKS, k3s, etc.)
• Default StorageClass (for persistent volumes)
• Network access to your S3 endpoint

Optional: Kata Containers

For stronger isolation, the executor can use Kata containers via RuntimeClass. Your cluster must have Kata installed and configured.

Credentials You Need

S3 Storage (required)

• Endpoint URL
• Region
• Access key ID
• Secret access key
• Bucket names (meta + content)

External Database (optional)

If not using internal CockroachDB:

• PostgreSQL/CockroachDB connection string