Kubernetes

Production-ready Kubernetes deployment using Helm with built-in observability.

Architecture

┌─────────────────────────────────────────────────────────────────────────────┐
│                           Kubernetes Cluster                                 │
├─────────────────────────────────────────────────────────────────────────────┤
│  Core Services (deployed by Helm):                                           │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────────┐                  │
│  │ CockroachDB │  │    Redis    │  │    API Service      │                  │
│  │  (3 nodes)  │  │ (job queue) │  │   (autoscaling)     │                  │
│  └─────────────┘  └─────────────┘  └─────────────────────┘                  │
│                                     ┌─────────────────────┐                  │
│                                     │   Executor Pool     │                  │
│                                     │   (autoscaling)     │                  │
│                                     └─────────────────────┘                  │
├─────────────────────────────────────────────────────────────────────────────┤
│  Observability Stack (optional):                                             │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────────┐                  │
│  │ Prometheus  │  │   Grafana   │  │       Tempo         │                  │
│  │  (metrics)  │  │ (dashboards)│  │    (tracing)        │                  │
│  └─────────────┘  └─────────────┘  └─────────────────────┘                  │
├─────────────────────────────────────────────────────────────────────────────┤
│  External (user provides):                                                   │
│  ┌─────────────────────────────────────────────────────────────────────┐    │
│  │                    S3-compatible Storage                             │    │
│  │           (AWS S3, Cloudflare R2, GCS, MinIO, etc.)                 │    │
│  └─────────────────────────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────────────────────────┘

Quick Start

Local Development (k3d)

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

Creates a complete local environment with monitoring in ~5 minutes.

→ Local Development Guide

Production

helm install flow-like ./helm -n flow-like --create-namespace \
  --set storage.external.endpoint='https://s3.example.com' \
  --set storage.external.accessKeyId='YOUR_KEY' \
  --set storage.external.secretAccessKey='YOUR_SECRET'

→ Production Installation Guide

What’s Included

Core Services

Component	Description
CockroachDB	3-node distributed SQL database
Redis	Job queue and execution state
API Service	Flow-Like API with autoscaling
Executor Pool	Reusable execution workers with autoscaling
DB Migration Job	Prisma migrations on install/upgrade

Observability Stack

Component	Description
Prometheus	Metrics collection from all services
Grafana	Pre-configured dashboards
Tempo	Distributed tracing with OpenTelemetry

Enable with --set monitoring.enabled=true (enabled by default).

Grafana Dashboards

Six pre-built dashboards for full visibility:

System Overview — Cluster-wide resource usage
API Service — Request rates, latencies, errors
Executor Pool — Job queue depth, execution metrics
CockroachDB — Query performance, replication
Redis — Commands/sec, memory usage
Tracing — Request traces via Tempo