onelab-k8s-1.27/gitops/README.md

# OneLab GitOps (Argo CD)

This directory is the **declarative source** for OneLab on Kubernetes. [Argo CD](https://argo-cd.readthedocs.io/) applies **two Helm-based sources** from Git (Argo invokes Helm; you do not use Helm CLI as the primary install path for the cluster).

This repository is **GitOps-only**: everything required to deploy OneLab on Kubernetes lives under **`gitops/`**.

## What gets deployed

| Helm release | Chart path | Namespace | Role |
|--------------|------------|-----------|------|
| `onelab` | [`charts/onelab`](charts/onelab) | `onelab` | Application workloads, Postgres, Redis, RabbitMQ, revproxy, ingress hooks |
| `onelab-obs` | [`observability/`](observability/) | `onelab` | Loki, Promtail, Grafana (logs) |

Both are wired by a single Argo CD [`Application`](argocd/application.yaml) using **`spec.sources`** (requires **Argo CD 2.6+**).

## Architecture

```mermaid
flowchart LR
  subgraph git [Git repository]
    V[gitops/values]
    C[charts/onelab]
    O[observability]
  end
  subgraph argo [Argo CD]
    App[Application onelab]
  end
  subgraph cluster [Kubernetes cluster]
    NS[namespace onelab]
  end
  V --> App
  C --> App
  O --> App
  App --> NS
```

1. You commit changes under **`gitops/values/`** (and optionally edit `repoURL` / `targetRevision` in the Application).
2. Argo reconciles: Helm renders `onelab` + `onelab-obs` into namespace **`onelab`**.
3. Sync waves order StatefulSets and app Deployments (Postgres → Redis/Rabbit/config → apps).

## Layout

| Path | Purpose |
|------|---------|
| [`charts/onelab`](charts/onelab) | OneLab Helm chart — **Argo source 1** |
| [`values/`](values/) | **Operator entry point**: [`values/env-example.yaml`](values/env-example.yaml), [`values/observability.yaml`](values/observability.yaml), templates — see [`values/README.md`](values/README.md) |
| [`observability/`](observability/) | Loki / Promtail / Grafana umbrella chart — **Argo source 2** (`releaseName: onelab-obs`) |
| [`argocd/application.yaml`](argocd/application.yaml) | `Application` manifest (`spec.sources`, destination namespace `onelab`) |
| [`argocd/jsonpatch-multisource.json`](argocd/jsonpatch-multisource.json) | One-time JSON patch if a live `Application` still has legacy `spec.source` only |

## Prerequisites

1. **Kubernetes** (e.g. k3s) with a default **StorageClass** for Postgres/Rabbit/Loki/Grafana PVCs (e.g. `local-path`).
2. **Container images** from `hub.andrewalliance.com` (or your mirror): registry credentials via Helm values or a pre-created `docker-registry` Secret — see [Private registry credentials](#private-registry-credentials).
3. **RabbitMQ TLS** before RabbitMQ starts: Kubernetes Secret `onelab-rabbit-tls`, or `rabbitmq.tls.embed` in private values — see [RabbitMQ TLS](#rabbitmq-tls).
4. **Host paths** when using `persistence.mode: hostPath`: `/opt/onelab/data` and `/opt/onelab/logs` on nodes that run those pods (aligned with Promtail in [`values/observability.yaml`](values/observability.yaml)), or use RWX storage for multi-node.

## Configuration (single place to edit)

- **OneLab app / ingress / DB / tokens / registry**: [`values/env-example.yaml`](values/env-example.yaml) — copy patterns, replace **`REPLACE_*`** placeholders, or overlay a gitignored file; see [`values/secrets.example.yaml`](values/secrets.example.yaml) and [`values/README.md`](values/README.md).
- **Observability**: [`values/observability.yaml`](values/observability.yaml) — Grafana password (**`REPLACE_GRAFANA_ADMIN_PASSWORD`**), ingress host, Promtail log hostPath.
- **Argo Git coordinates**: [`argocd/application.yaml`](argocd/application.yaml) — set `repoURL`, `targetRevision`, and `helm.valueFiles` if you add e.g. `secrets.local.yaml`.

The chart default [`charts/onelab/values.yaml`](charts/onelab/values.yaml) holds non-secret structure and safe placeholders only; **do not rely on it for production secrets**.

## Bootstrap checklist

1. Fork or clone this repository and push it to a Git remote your cluster can reach.
2. Edit [`argocd/application.yaml`](argocd/application.yaml): **`repoURL`** → your remote, **`targetRevision`** → your branch/tag.
3. Edit [`values/env-example.yaml`](values/env-example.yaml): DNS hostnames, TLS secret names, cert-manager issuer, and all **`REPLACE_*`** values (or use a gitignored overlay).
4. Edit [`values/observability.yaml`](values/observability.yaml): Grafana host/password aligned with your DNS and TLS.
5. If the Git repo is **private**, register it in Argo CD (see below).
6. Ensure **RabbitMQ TLS** Secret exists (or use embedded TLS in private values).
7. Apply the Application:

   ```bash
   kubectl apply -f gitops/argocd/application.yaml
   ```

8. In Argo UI or CLI, confirm sync to namespace **`onelab`** and fix any `ImagePullBackOff` / TLS issues using the sections below.

**Single controller:** Use **only** this Argo CD `Application` for `onelab` / `onelab-obs`. Do not manage the same namespace with a parallel **Helm CLI** release.

## Private registry credentials

Set `registry.username` / `registry.password` in values (prefer a **gitignored** file merged last), or create the Secret manually:

```bash
kubectl create secret docker-registry hub-andrewalliance -n onelab \
  --docker-server=hub.andrewalliance.com \
  --docker-username='YOUR_USER' \
  --docker-password='YOUR_PASSWORD'
```

…then set `registry.createPullSecret: false` and keep `imagePullSecrets: [{ name: hub-andrewalliance }]` in values.

### StatefulSet pods still get `401 Unauthorized` / `ImagePullBackOff` after enabling registry auth

If `db-0` / `rabbitmq-0` were created **before** `imagePullSecrets` existed, delete those pods once so they pick up the new Pod spec:

```bash
kubectl delete pod -n onelab db-0 rabbitmq-0
```

## Argo CD private Git repository

If the Application shows `authentication required: Unauthorized`, register the repo (use a deploy token or PAT with read access):

```bash
argocd repo add https://github.com/YOUR_ORG/YOUR_REPO.git \
  --username git \
  --password YOUR_TOKEN
```

## RabbitMQ TLS

Secret **`onelab-rabbit-tls`** must exist in namespace **`onelab`** before RabbitMQ starts (keys: typically `tls.crt`, `tls.key`, and optionally chain). **PEM material is not shipped in this GitOps tree** — generate certs or use your PKI, then:

```bash
kubectl create secret tls onelab-rabbit-tls -n onelab \
  --cert=path/to/fullchain.pem \
  --key=path/to/key.pem
```

Alternatively, set `rabbitmq.tls.embed: true` and supply `rabbitmq.tls.crt` / `rabbitmq.tls.key` / `rabbitmq.tls.fullchain` via a **private** values file (never commit real keys).

## Deploy with Argo CD (details)

**Requirements:** Argo CD **2.6+** (`spec.sources`).

Each entry under `spec.sources` has its own `helm.releaseName` and `helm.valueFiles` (paths are **relative to that source’s `path`**):

- Source `gitops/charts/onelab` → e.g. `../../values/env-example.yaml`, optionally `../../values/secrets.local.yaml`
- Source `gitops/observability` → e.g. `../../values/observability.yaml`

### Migrating `spec.source` → `spec.sources`

If the `onelab` `Application` was created earlier with **`spec.source` only**, a plain `kubectl apply` may **not** remove `spec.source`, and Argo will not reconcile the observability chart.

Check:

```bash
kubectl get application onelab -n argocd -o jsonpath='{.spec.source}{"\n"}{.spec.sources}{"\n"}'
```

If `source` is set and `sources` is empty, patch once (edit `repoURL` in the patch file to match your remote):

```bash
kubectl patch application onelab -n argocd --type json --patch-file gitops/argocd/jsonpatch-multisource.json
```

## Observability (Loki / Promtail / Grafana)

The umbrella chart under [`observability/`](observability/) deploys:

- **Loki** — log storage (SingleBinary, filesystem PVC; retention from [`values/observability.yaml`](values/observability.yaml)).
- **Promtail** — DaemonSet: Kubernetes pod logs plus **OneLab file logs** from the host path configured in [`values/observability.yaml`](values/observability.yaml) (keep in sync with OneLab `persistence.hostPath.logs`).
- **Grafana** — Explore; datasource points at this release’s Loki gateway.

### First-time setup

1. Set a strong **`grafana.adminPassword`** in [`values/observability.yaml`](values/observability.yaml) (or use `admin.existingSecret` per the upstream Grafana chart).
2. Align **Grafana ingress host** with `grafana.ini.server.domain` / `root_url` in the same file.
3. **Multi-node** — with `hostPath` logs, each node only sees its own files; Promtail runs on every node.

### OneLab-only ingestion

Promtail **`extraRelabelConfigs`** keep only pods in namespace **`onelab`**. Host file logs are tagged **`namespace: onelab`**, **`component: host-logs`**.

### Dashboard: **OneLab logs**

Grafana sidecar loads the dashboard from `observability/dashboards/onelab-logs.json` (`uid` `onelab-logs`).

#### Grafana pod: `init-chown-data` CrashLoopBackOff

This repo sets **`grafana.initChownData.enabled: false`** with **`fsGroup: 472`** for Pod Security–friendly clusters. If Grafana cannot write to the PVC, delete the Grafana PVC once after changing values or adjust Pod Security for namespace `onelab`.

### Access Grafana

Ingress **`grafana-onelab`** is defined in `observability/templates/ingress-grafana-onelab.yaml`. Defaults use example hosts in [`values/observability.yaml`](values/observability.yaml) (`grafana.onelab.example.com`); change to your DNS and TLS Secret name.

```bash
kubectl -n onelab port-forward svc/onelab-obs-grafana 3000:80
```

### Maintainers: vendored chart dependencies

From `gitops/observability/`:

```bash
helm dependency update
```

Commit updated `Chart.lock` and `charts/*.tgz` so Argo can render without live Helm repo access at sync time.

## Application configuration (`configurations.yml`)

For Kubernetes you do **not** need a separate `configurations.yml` in Git. The OneLab chart renders it from [`charts/onelab/files/configurations.gotmpl`](charts/onelab/files/configurations.gotmpl) into Secret **`onelab-configurations`**.

1. **Values (recommended)** — set `onelab.compliance`, `onelab.ldap`, etc. See [`values/instance-overrides.example.yaml`](values/instance-overrides.example.yaml) and add paths under **`spec.sources[].helm.valueFiles`** for the `gitops/charts/onelab` source.
2. **Bring your own Secret** — set `configuration.existingSecretName`; the Secret must contain key **`configurations.yml`**.

## Ingress (web UI)

Set `ingress.enabled`, `ingress.host`, and optional TLS in [`values/env-example.yaml`](values/env-example.yaml). Traffic goes to Service **`revproxy`**. On k3s, `ingress.className: traefik` matches the default controller. For cert-manager, set `ingress.certManager.clusterIssuer` and TLS secret name; DNS for `ingress.host` must resolve before ACME completes.

## Security notes for public repositories

- Never commit real passwords, tokens, or TLS private keys. Use **`REPLACE_*`** in tracked files and a **`*.local.yaml`** overlay (ignored at repo root) for secrets.
- If this repo ever contained real credentials, **rotate** them after sanitizing Git history or publishing a clean fork.

## kubectl / credentials

If `kubectl` reports *You must be logged in*, refresh your kubeconfig before applying manifests.

## Developer note (local render)

Running **`helm template` on Windows** against some paths can return empty `.Files.Get` content; the OneLab chart uses `fromYaml (.Files.AsConfig)` where needed. **Argo CD runs on Linux** and renders the same charts in-cluster.

## Not covered by this chart

- **Edge proxy** in front of the cluster — use **Ingress** + `revproxy` and optional cert-manager.
- **Non-Kubernetes** install paths — not included; use Kubernetes Secrets or external secret operators as needed.