Install Grown Workspace

What gets installed

The chart is self-contained — it bundles everything Grown needs so a fresh cluster becomes a working workspace:

• grown — the platform (all the apps: Drive, Mail, Calendar, Docs, Sheets, Games, …).
• PostgreSQL — a plain StatefulSet (no operator required); swap in CNPG or a managed DB for production.
• MinIO — S3-compatible object storage for files, photos, video and music (swap in rustfs/any S3).
• Zitadel — OIDC single sign-on, with the Grown OIDC app auto-provisioned on install.

Everything is plain Kubernetes objects, arm64-friendly, with sensible defaults that come up out of the box on kind.

Method 1 — Helm chart recommended

The simplest path. Requires a Kubernetes cluster and helm.

# clone the repo (the chart lives in deploy/helm/grown)
git clone https://github.com/grown-platform/grown.git
cd grown

# install everything into the "grown" namespace
helm install grown deploy/helm/grown \
  -n grown --create-namespace \
  --set domain=grown.example.com

That's it — Postgres, MinIO, Zitadel (with the OIDC app provisioned) and grown all come up. Watch it with kubectl -n grown get pods -w; grown is ready when /healthz returns 200. The default admin is printed in the install notes.

Common values

See values.yaml for the full list, and the chart README for production guidance.

Method 2 — Raw manifests (kubectl apply)

No Helm? Apply the pre-rendered all-in-one manifest.

# edit the hostname/values in the file first, then:
kubectl create namespace grown
kubectl apply -n grown -f deploy/manifests/grown.yaml

deploy/manifests/grown.yaml is rendered from the same chart. To customise it, re-render with your own values:

helm template grown deploy/helm/grown -n grown \
  --set domain=grown.example.com > my-grown.yaml
kubectl apply -n grown -f my-grown.yaml

Method 3 — Local machine with kind

Try the whole platform on your laptop in a throwaway cluster. Requires kind, kubectl and helm (Docker/Podman under the hood). This is the exact path the chart is smoke-tested on.

kind create cluster --name grown

helm install grown deploy/helm/grown \
  -n grown --create-namespace \
  --set domain=grown.local --set ingress.type=none

# wait for grown, then reach it via a port-forward
kubectl -n grown rollout status deploy/grown
kubectl -n grown port-forward svc/grown 8080:8080
# open http://localhost:8080/healthz  -> 200

Method 4 — Raspberry Pi 5 (16 GB) cluster

Grown runs well on a small arm64 Pi 5 cluster — the app itself is a lightweight Go binary. Recommended: 3–4× Pi 5 (16 GB) with NVMe (the M.2 HAT), not SD cards.

• arm64 images. grown and the bundled Postgres/MinIO/Zitadel all have arm64 images. Pin image.tag to an arm64-built grown tag.
• Real storage. Use NVMe + a storage class like local-path or Longhorn for Postgres I/O and S3 blobs — set --set storageClass=longhorn. Avoid SD cards.
• No heavyweight operators. The chart needs none — ideal for Pi. (Skip Ceph; use Longhorn/local-path.)
• Headroom. Core grown (app + Postgres + Zitadel + MinIO) is ~1.5–2.5 GiB — comfortable. Heavy extras (transcoding, live streaming, AV mail scanning) are what strain Pi cores; keep those modest.

helm install grown deploy/helm/grown \
  -n grown --create-namespace \
  --set domain=grown.example.com \
  --set storageClass=longhorn \
  --set image.tag=<arm64-tag>

A single 16 GB Pi 5 can host core grown for a household; a 3–4 node cluster comfortably runs grown plus heavier add-ons (mail, photos) for a small team.