Art-gallery/Documentation/environments.md

# Development and production environments

Gallery uses **one PostgreSQL server** on TrueNAS (`192.168.10.122`) with **two databases**. The **dev PC** (`192.168.10.70`) runs `npm run dev:web` on port **5173**, published at **`https://devgallery.mysuperlab.netcraze.pro`** via Keenetic.

| Environment | Database | Public URL | App host |
|-------------|----------|------------|----------|
| **Development** | `gallery_dev` | `https://devgallery.mysuperlab.netcraze.pro` | Dev PC `192.168.10.70:5173` |
| **Production** | `gallery_prod` | `https://gallery.mysuperlab.netcraze.pro` | TrueNAS container `192.168.10.122:5173` |

Version tracking:
- Runtime version endpoint: `/api/version` (reports `app_env`, `app_version`, `image_tag`, `git_sha`, `built_at`, `db_name`)
- `npm run prod:docker:publish` auto-creates a release tag like `20260709-155412-f72ddcc` (build time + git commit), pushes it and `latest`
- Version metadata is baked into the Docker image; TrueNAS only needs `gallery-web:latest` with `pull_policy: always`

---

## Development-first workflow (default)

**All day-to-day work happens on dev.** Code edits, database changes, catalog updates, image fixes, migrations, and testing use **`gallery_dev`**, the repo’s **`data/images/`**, and **https://devgallery.mysuperlab.netcraze.pro** (or `http://localhost:5173` on the dev PC).

Production is **not** updated on every change. Promote to prod on a **regular cadence** (typically about once a week) or when you explicitly decide to release:

| Phase | Where | What |
|-------|-------|------|
| **Daily** | Dev PC + `gallery_dev` | Feature work, fixes, seed/migrate, curator edits, image pipeline |
| **Release** | Dev PC → TrueNAS | Backup dev → restore prod DB (if data changed) → sync images → `docker:publish` → restart **gallery-web** |

Unless a task explicitly says **prod** or **production**, assume the target is **dev** — including when asking an assistant to change, fix, or modify code, data, or the database.

**Never** point the dev PC `.env` at `gallery_prod`. Prod-only files (`infra/docker/.env.prod`, `db:restore:prod`, `images:sync-to-prod`) are for scheduled releases, not routine development.

---

## Where to run what (quick reference)

| Step | Machine | Interface | Privilege | What |
|------|---------|-----------|-----------|------|
| Stop dev servers | Dev PC `192.168.10.70` | PowerShell (normal) | your user | Close `dev:web` / `dev:server` / `dev:client` terminals |
| **DB split (one-time)** | Dev PC | **pgAdmin** → Query Tool on `postgres` | **postgres superuser** | [`db/split-dev-prod-pgadmin.sql`](../db/split-dev-prod-pgadmin.sql) step by step |
| Dev migrate / dev:web | Dev PC | PowerShell (normal) | your user | `npm run dev:migrate`, `npm run dev:web` |
| Image dir on TrueNAS | TrueNAS `192.168.10.122` | **Shell** (SSH or UI → System Settings → Shell) | root / sudo | `bash infra/docker/truenas-setup.sh` |
| Copy images to prod | Dev PC | PowerShell (normal) | SMB `\\192.168.10.122\Gallery` | `net use` then `npm run devtoprod:images` |
| Build + push Docker image | Dev PC | **PowerShell as Administrator** | admin (for LAN hosts entry) | `npm run prod:docker:publish` |
| Install prod app | TrueNAS | **Web UI** → Apps → Custom App | admin | Paste `infra/docker/compose.truenas.yaml` |
| Verify prod | Dev PC or TrueNAS | PowerShell / browser | any | `curl.exe -sk https://gallery.mysuperlab.netcraze.pro/api/bounds` |

You do **not** need `psql` on TrueNAS. Database work is done from **pgAdmin on the dev PC** connected to `192.168.10.122:5432`.

---

## One-time setup (full walkthrough)

### Step A — Stop Gallery on the dev PC

**Where:** Dev PC — PowerShell or open terminal tabs (no admin needed)

Close anything using ports **5173**, **3451**, or **3520**:

- Stop `npm run dev:web`, `npm run dev:server`, `npm run dev:client`
- If prod container already runs on TrueNAS: **TrueNAS Web UI** → Apps → **gallery-web** → **Stop**

### Step B — Split the database (pgAdmin)

**Where:** Dev PC — **pgAdmin** (not TrueNAS shell)

1. Open **pgAdmin**.
2. Add/connect to server:
   - **Host:** `192.168.10.122`
   - **Port:** `5432`
   - **Maintenance database:** `postgres`
   - **Username:** `postgres` (superuser — **not** `gallery`)
   - **Password:** your postgres password
3. Tree: **Servers** → your server → **Databases** → click **`postgres`**
4. **Tools** → **Query Tool** (or right-click `postgres` → Query Tool)
5. **File** → **Open** → `Gallery\db\split-dev-prod-pgadmin.sql`
6. Run **each STEP separately** (highlight from `-- STEP N` through that section, press **F5** / Execute):

   | Step | Action |
   |------|--------|
   | **STEP 0** | Pre-flight — should show database `Gallery` |
   | **STEP 1** | Terminate connections |
   | **STEP 2** | `ALTER DATABASE "Gallery" RENAME TO gallery_prod` (use lowercase variant only if STEP 0 showed `gallery`) |
   | **STEP 3** | `CREATE DATABASE gallery_dev WITH TEMPLATE gallery_prod` (wait ~1–2 min) |
   | **STEP 4** | `GRANT` to user `gallery` |
   | **STEP 5** | Verify — should show `gallery_prod` and `gallery_dev` |
   | **STEP 6** | **New Query Tool on `gallery_dev`** (not postgres!) — `SELECT current_database()` then `SELECT count(*) FROM paintings` |

If STEP 2 fails with “database already exists”, STEP 0 likely already shows `gallery_prod` — skip STEP 2 and continue from STEP 3 (skip STEP 3 too if `gallery_dev` exists).

**Alternative (dev PC with psql installed):**

```powershell
cd T:\Repo\Gallery
psql -h 192.168.10.122 -U postgres -d postgres -f db/split-dev-prod.sql
```

**Alternative (Node, postgres password in env):**

```powershell
cd T:\Repo\Gallery
$env:PGHOST="192.168.10.122"; $env:PGUSER="postgres"; $env:PGPASSWORD="YOUR_POSTGRES_PASSWORD"
npm run infra:db:split-dev-prod
```

### Step C — Configure dev PC and test

**Where:** Dev PC — PowerShell (normal), repo root

1. Edit `.env` (copy from `.env.example` if needed):

   ```env
   DB_NAME=gallery_dev
   PORT=3451
   PUBLIC_URL=https://devgallery.mysuperlab.netcraze.pro
   SESSION_SECRET=your-long-random-secret
   CURATOR_USERNAME=curator
   CURATOR_PASSWORD=your-secure-password
   ```

   Omit `SESSION_COOKIE_SECURE` so cookies follow the request scheme (`TRUST_PROXY` + HTTPS → Secure). Set `true`/`false` to force. `npm run dev:migrate` creates auth tables/roles and bootstraps the first **admin** when `users` is empty. Reset that account later with `npm run dev:reset-curator`. Create additional staff via the in-app **Users** page. Admins browse curator actions on **Activity** (`/api/audit`), which always reads the DB named by `DB_NAME` for that environment (`gallery_dev` here; `gallery_prod` on prod). Audit history is not synced by harmonize/devtoprod.

2. Run:

   ```powershell
   cd T:\Repo\Gallery
   npm run dev:migrate
   npm run dev:web
   ```

3. Open **https://devgallery.mysuperlab.netcraze.pro** (or http://localhost:5173 on the dev PC)

### Step D — Prepare prod image folder on TrueNAS

**Where:** TrueNAS — **Shell** (SSH to `192.168.10.122`, or TrueNAS UI → **System Settings** → **Shell**)

```bash
# Copy script to TrueNAS first, or paste commands manually:
mkdir -p /mnt/BasePool/Applications/Gallery/data/images/portraits
mkdir -p /mnt/BasePool/Applications/Gallery/data/images/paintings/thumbs
chown -R 1001:1001 /mnt/BasePool/Applications/Gallery
chmod -R u+rwX,g+rwX /mnt/BasePool/Applications/Gallery
```

Or from a checkout on TrueNAS: `bash infra/docker/truenas-setup.sh`

### Step E — Copy images dev → prod volume

**Where:** Dev PC — PowerShell (normal), repo root

Requires SMB share **`Gallery`** → `/mnt/BasePool/Applications/Gallery` on TrueNAS.

**Connect to the share first** (once per Windows session), then sync:

```powershell
# Map share (use your TrueNAS SMB user/password)
net use \\192.168.10.122\Gallery /user:YOUR_TRUENAS_USER

cd T:\Repo\Gallery
npm run devtoprod:images
```

UNC destination: `\\192.168.10.122\Gallery\data\images`

Type `yes` when prompted. First run copies ~1000+ files (several minutes).

If `net use` fails, open `\\192.168.10.122\Gallery` in File Explorer and sign in, then retry.

### Step F — Build and push Docker image

**Where:** Dev PC — **PowerShell as Administrator** (for fast LAN push to Gitea)

Prerequisites: **Docker Desktop running**, logged in to Gitea.

```powershell
cd T:\Repo\Gallery
docker login gitea.mysuperlab.netcraze.pro
npm run prod:docker:publish
```

Use normal PowerShell with `-SkipHosts` if you already added `192.168.10.122 gitea.mysuperlab.netcraze.pro` to `C:\Windows\System32\drivers\etc\hosts`.

### Step G — Install production on TrueNAS

**Where:** TrueNAS — **Web UI** (browser)

1. **Apps** → **Discover Apps** → **Custom App** → **Install via Docker Compose**
2. Paste contents of `infra/docker/compose.truenas.yaml` from the repo
3. Replace `YOUR_POSTGRES_PASSWORD` with the `gallery` user password
4. Replace `REPLACE_WITH_LONG_RANDOM_SECRET` and `REPLACE_WITH_SECURE_PASSWORD` for `SESSION_SECRET` and `CURATOR_PASSWORD`
5. **Apps** → **Settings** → register Gitea registry (`gitea.mysuperlab.netcraze.pro`, token with `read:package`)
6. Deploy → wait for **gallery-web** to show **Running**
7. Run `npm run dev:migrate` against `gallery_prod` if auth tables are not yet applied (or migrate from dev PC with prod env)

### Step H — Verify production

**Where:** Dev PC — PowerShell (normal) or any browser

```powershell
curl.exe -sk https://gallery.mysuperlab.netcraze.pro/api/bounds
curl.exe -s http://192.168.10.122:5173/api/bounds
```

Open **https://gallery.mysuperlab.netcraze.pro/** — timeline and sample painting images should load. Click an artist portrait or movement label to enter a 3D hall. **Curator login** (top-right) unlocks debug mode and Checkup.

---

## Environment files

| File | Git | Purpose |
|------|-----|---------|
| [`.env`](../.env) | ignored | **Dev** — `DB_NAME=gallery_dev`, `PORT=3451` |
| [`.env.example`](../.env.example) | tracked | Dev template |
| [`infra/docker/.env.prod`](../infra/docker/.env.prod) | ignored | **Prod** — migrate/restore prod scripts |
| [`infra/docker/.env.prod.example`](../infra/docker/.env.prod.example) | tracked | Prod template |
| [`infra/docker/compose.truenas.yaml`](../infra/docker/compose.truenas.yaml) | tracked | TrueNAS Custom App |

Never point dev `.env` at `gallery_prod`. Prod scripts refuse dev env files.

---

## Dev public access (Keenetic)

**Where:** Keenetic router Web UI (not dev PC)

| Field | Value |
|-------|-------|
| Domain | `devgallery.mysuperlab.netcraze.pro` |
| Upstream IP | `192.168.10.70` |
| Upstream port | **`5173`** |
| **Protocol to device** | **`http`** (not https) |
| Preserve Host | ON |

Vite on the dev PC speaks **plain HTTP** only. Keenetic terminates HTTPS from the browser, then must forward **HTTP** to `192.168.10.70:5173`.

If **Protocol to device** is `https`, Keenetic tries TLS against Vite → **502 Bad Gateway** (`Server: Web server`).

**Verify from dev PC** (servers must be running: `npm run dev:web`):

```powershell
# PowerShell: curl is an alias — use curl.exe for -k, or Invoke-WebRequest
curl.exe -s -o NUL -w "HTTP %{http_code}`n" http://192.168.10.70:5173/
curl.exe -sk -o NUL -w "HTTP %{http_code}`n" https://devgallery.mysuperlab.netcraze.pro/

# Or native PowerShell (skip cert check):
Invoke-WebRequest -Uri https://devgallery.mysuperlab.netcraze.pro/ -SkipCertificateCheck | Select-Object StatusCode
```

Expect **HTTP 200** (not 502).

**`.env`:** use the same scheme as the browser URL. If Keenetic serves HTTPS publicly:

```env
PUBLIC_URL=https://devgallery.mysuperlab.netcraze.pro
TRUST_PROXY=true
SESSION_SECRET=your-long-random-secret
```

Omit `SESSION_COOKIE_SECURE` for auto Secure cookies behind Keenetic HTTPS. Prod (`infra/docker/.env.prod`): set `SESSION_COOKIE_SECURE=true` and the same `SESSION_SECRET` / `CURATOR_*` vars on the TrueNAS app environment.

Restart `npm run dev:web` after changing `PUBLIC_URL`.

---

## Production public access (Keenetic)

**Where:** Keenetic router Web UI

| Field | Value |
|-------|-------|
| Domain | `gallery.mysuperlab.netcraze.pro` |
| Upstream IP | **`192.168.10.122`** (TrueNAS — not the dev PC) |
| Upstream port | **`5173`** |
| **Protocol to device** | **`http`** (container speaks HTTP, not HTTPS) |
| Preserve Host | ON |

The prod container listens on plain HTTP on port **5173**. Keenetic terminates HTTPS from the browser and must forward **HTTP** to TrueNAS.

**Verify the app first (bypass Keenetic):**

```powershell
curl.exe -s http://192.168.10.122:5173/api/bounds
```

Expect JSON immediately (~200). If this works but the public URL returns **502/504**, the Keenetic rule is wrong (IP, port, or `https` to device).

**Then verify public URL:**

```powershell
curl.exe -sk https://gallery.mysuperlab.netcraze.pro/api/bounds
```

See also [Drunkmeyou gitea-https-keenetic-npm-setup.md](../../Drunkmeyou/Documentation/gitea-https-keenetic-npm-setup.md) — same Keenetic HTTPS → HTTP upstream pattern.

---

## Daily development

**Where:** Dev PC — PowerShell (normal), repo root

| Task | Command |
|------|---------|
| Public dev URL (Keenetic) | `npm run dev:web` |
| Fast local HMR (no Keenetic) | `npm run dev:server` + `npm run dev:client` |
| Refresh dev DB from prod | `npm run prodto:dev:db` |
| Pull prod images to dev | `npm run prodto:dev:images` |
| Merge dev ↔ prod catalog + images (incremental) | `npm run harmonize` — see [harmonize-dev-prod.md](harmonize-dev-prod.md) |

---

## Promote dev → prod (scheduled release)

Run this when you are ready to ship dev to production — **not** after every small change. Typical cadence: **about once a week**.

For **incremental** dev ↔ prod merge (both sides edited), use [harmonize-dev-prod.md](harmonize-dev-prod.md) instead of full restore.

**One command (recommended):** copy [`infra/deploy/devtoprod.config.example.json`](../infra/deploy/devtoprod.config.example.json) to `infra/deploy/devtoprod.config.json`, edit it, then `npm run devtoprod:release` or `deploy-dev-to-prod.cmd`. See [deploy-dev-to-prod.md → One-command release](deploy-dev-to-prod.md#one-command-release-automated).

**Manual steps:** detailed runbook (per-change decision matrix, schema migration, rollback): [deploy-dev-to-prod.md](deploy-dev-to-prod.md).

**Where:** Dev PC unless noted

1. Finish and test on https://devgallery.mysuperlab.netcraze.pro
2. `npm run devtoprod:thumbnails` — skip if no painting/portrait images changed
3. `npm run dev:db:backup`
4. `npm run devtoprod:db:restore -- --file db/DataBackup/gallery_dev_data_....txt` (type `yes`) — skip if only code changed and prod DB should stay as-is
5. `npm run devtoprod:images` — skip if no new/changed images
6. `npm run prod:docker:publish` — required when application code changed
7. **TrueNAS Web UI** → restart **gallery-web**
8. Verify https://gallery.mysuperlab.netcraze.pro

---

## Database scripts

| Command | Where | Purpose |
|---------|-------|---------|
| pgAdmin + `split-dev-prod-pgadmin.sql` | Dev PC pgAdmin | One-time split (recommended) |
| `npm run infra:db:split-dev-prod` | Dev PC PowerShell | Same split (needs `PGUSER=postgres`) |
| `npm run prodto:dev:db` | Dev PC PowerShell | Clone prod → dev |
| `npm run dev:db:backup` | Dev PC PowerShell | Dev backup |
| `npm run devtoprod:db:restore` | Dev PC PowerShell | Restore into prod |
| `npm run harmonize` | Dev PC PowerShell | Bidirectional catalog + image merge — [harmonize-dev-prod.md](harmonize-dev-prod.md) |
| `npm run harmonize:db` | Dev PC PowerShell | DB merge only |
| `npm run harmonize:images` | Dev PC PowerShell | Image merge + artists/paintings checkup/path sync + regenerate thumbs on both sides |

## Image sync

| Command | Where | Direction |
|---------|-------|-----------|
| `npm run harmonize` | Dev PC PowerShell | Bidirectional merge (mtime newer wins) — [harmonize-dev-prod.md](harmonize-dev-prod.md) |
| `npm run devtoprod:release` | Dev PC PowerShell | Full config-driven promote (see [deploy-dev-to-prod.md](deploy-dev-to-prod.md#one-command-release-automated)) |
| `npm run devtoprod:thumbnails` | Dev PC PowerShell | Rebuild painting + portrait thumbs on dev before promote |
| `npm run devtoprod:images` | Dev PC PowerShell | Dev → TrueNAS volume |
| `npm run prodto:dev:images` | Dev PC PowerShell | TrueNAS → dev repo |

## Safety guards

- Prod restore reads only `infra/docker/.env.prod`
- Dev backup refuses `_prod` database names without `--prod`
- Destructive prod ops require typing `yes` or `CONFIRM_PROD=1`

## Legacy deployment

Node on dev PC `:3520` + nginx → Vite `:5173` is optional; see [`deploy/nginx-gallery.conf`](../deploy/nginx-gallery.conf). Production should use TrueNAS Docker on `:5173`.

Deploy details: [`infra/docker/DEPLOY-truenas.md`](../infra/docker/DEPLOY-truenas.md).