---
name: memorains-deploy
description: Deploy memorains to production servers. Use when the user wants to deploy, ship, push, or release memorains, or mentions putting it on the server/VPS/production.
---

# Memorains Deploy

Deploy the memorains project to one or more production servers.

## Step 0: Determine where to deploy

**If the user already specified which host(s) to deploy to**, use them.

**Otherwise, ask:** "Which server(s) should I deploy to?" — and wait. Never infer
a host from `~/.ssh/config`, shell history, README files, or past sessions.

## Prerequisites

- SSH access to the target servers must work (key-based auth)
- Servers must have `podman` and `podman-compose` installed

## Preferred: deploy with the script

The companion script `deploy_web.sh` (in this skill directory) automates the
whole flow: it builds the package, discovers each host's project root from the
running `reno_note_nodejs` container mount, uploads, restarts the containers,
and verifies the result. Hosts are always passed as arguments — the script
contains no host information.

Run it from the repository root (walk up from `.agents/skills/memorains-deploy/`
to the repository root, i.e. `<PROJECT_ROOT>`):

```bash
bash <PROJECT_ROOT>/.agents/skills/memorains-deploy/deploy_web.sh -y <HOST> [HOST...]
```

Useful flags:

| Flag | Effect |
|------|--------|
| `-y`, `--yes` | Skip the confirmation prompt |
| `--no-build` | Reuse the existing `script/out/package.tar.gz` |
| `--no-verify` | Skip the post-deploy container/HTTP checks |

Example:

```bash
bash <PROJECT_ROOT>/.agents/skills/memorains-deploy/deploy_web.sh -y example.com note.example.com
```

If the script exits non-zero — or a host's containers are already down so its
project root cannot be discovered (in that case export `DEPLOY_PROJECT_ROOT` or
fall back to the manual steps) — continue with the manual steps below.

## Manual steps (fallback)

Execute these steps in order for each server. Wait for each step to complete
before starting the next.

For each host, discover the project root and package directory by inspecting the
running containers. Run this on the server:

```bash
ssh <HOST> "podman inspect reno_note_nodejs --format '{{range .Mounts}}{{if eq .Destination \"/app/server\"}}{{.Source}}{{end}}{{end}}'"
```

This returns the mount source path like `<PROJECT_ROOT>/package/server`. Strip
the trailing `/server` to get the package directory, and strip `/package` to get
the project root where `package.tar.gz` should be uploaded.

For example, `/srv/app/note/package/server` → package dir:
`/srv/app/note/package`, project root: `/srv/app/note`.

If no containers are running, ask the user for the project root path.

### 1. Build the web package locally

```bash
cd <PROJECT_ROOT>/script && bash build_web_package.sh
```

This produces `out/package.tar.gz`.

### 2. Upload to the server

```bash
scp <PROJECT_ROOT>/script/out/package.tar.gz <HOST>:<PROJECT_ROOT>/
```

### 3. Stop and remove existing containers

```bash
ssh <HOST> "cd <PACKAGE_DIR> && podman-compose down 2>&1 || true"
```

The `|| true` handles the case where no containers are currently running.

### 4. Clean and extract the new package

```bash
ssh <HOST> "rm -rf <PACKAGE_DIR> && cd <PROJECT_ROOT> && tar -zxf package.tar.gz"
```

### 5. Start the new containers

```bash
ssh <HOST> "cd <PACKAGE_DIR> && podman-compose up -d"
```

### 6. Verify deployment

```bash
ssh <HOST> "podman ps --filter name=reno_note --format '{{.Names}} {{.Status}}'"
```

All three containers (`reno_note_mariadb`, `reno_note_nodejs`, `reno_note_nginx`)
should show as **Up**.

## What gets deployed

| Container | Purpose |
|-----------|---------|
| `reno_note_mariadb` | MariaDB database (data persisted via named volume) |
| `reno_note_nodejs` | Express + WebSocket server (Node.js Alpine) |
| `reno_note_nginx` | Nginx reverse proxy (ports 80/443) |