verae-ops/GETTING-STARTED.md

131 lines
5.8 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Bring Verae Time × Zapier online
This is the **top document** for install, first-up, and monitoring. Platform-specific detail lives in the numbered ops chapters; do not start there.
**Catalog:** https://zapier.georgelambert.org/packages/verae-ops/GETTING-STARTED.pdf
**Billing / users (zappier-edge):** https://zapier.georgelambert.org/packages/zappier/docs/USER-MANUAL.pdf
**Operator console:** http://127.0.0.1:3850/ (loopback) · https://zapier.georgelambert.org/packages/verae-fleet/docs/CONSOLE.pdf
## What you are turning on
| Piece | Job | Default |
|-------|-----|---------|
| **access-authz** | NATS authorization step for every access plane | `:3020` |
| **access-web / api / leaf / zapier / staff** | Public HTTP doors | `:3021:3025` |
| **identity** | `veraeUserId` bind/lookup | `:3026` |
| **zappier-edge** | Identity cache, Stripe, admin (loopback only) | `127.0.0.1:3000` (lab; NS1 all-in-one uses `:13000` if `:3000` is taken) |
| **staff-iam** | Named staff users, roles, sessions | `:3028` |
| **staff-session** | Staff cookie (proxies IAM) | `:3027` |
| **verae-keep** | Restart crashed units unless admin paused/stopped | keep `:3860`, watch `:3861` |
| **jobs-events** | Mailbox `verae.zapier.jobs.events` | `:3030` |
| **account-balance** | NATS prepaid balances (SoT) | `:3010` |
| **customer-service** | CS credit additions + review | `:3011` |
| **sales-pricing** | Per-customer sales quotes | `:3012` |
| **accounting-export** | QuickBooks IIF / CSV | `:3013` |
| **verae-middleware** | Zapier `/zapier/v1`, chain, NATS | `:3100` |
| **NATS JetStream** | Private jobs / archive | `:4222` **not public** |
| **worm ×3, tree-node ×3** | Off-chain files and Merkle leaves | fleet floor |
| **fleet** | Replica floors, SSH hosts, Trace, Docs | `:3850` on `0.0.0.0` |
Zapier cloud talks **only** to **access-zapier** (`:3024`) over HTTPS. Browsers use access-web. Customer API keys use access-api. Staff use access-staff. zappier-edge is loopback.
The billing and user platform is the standalone **zappier** project (`/Users/marchon/zappier`, Forgejo [zappier-edge](https://git.georgelambert.org/marchon/zappier-edge)): portal signup, TOTP, API keys, rate card, Stripe meter, PO invoices, admin users. Verae middleware does **not** replace it.
## 15-minute lab (one machine)
NS1 is Proxmox. Extra worm/tree LXC: [verae-proxmox](https://git.georgelambert.org/marchon/verae-proxmox) `create-worker.sh` (VMID 510, `10.10.10.20`). Off-box watch: [verae-uptime](https://git.georgelambert.org/marchon/verae-uptime). Backup: [verae-backup](https://git.georgelambert.org/marchon/verae-backup). Tagged upgrade: [verae-deploy](https://git.georgelambert.org/marchon/verae-deploy).
Bare metal / VM (Debian or Alpine): install OS + Node first, then clone.
```bash
git clone ssh://git@git.georgelambert.org:2223/marchon/verae-bootstrap.git
cd verae-bootstrap
bash scripts/host-deps.sh --type ns1-all-in-one
export VERAE_SRC=$HOME/verae-src
bash scripts/fetch.sh ns1-all-in-one
```
Docker lab from the monorepo:
```bash
git clone ssh://git@git.georgelambert.org:2223/marchon/master-zapier-plan-draft.git
cd master-zapier-plan-draft
export MONOREPO="$PWD"
cd packages/verae-ops
docker compose up --build
```
Check:
```bash
curl -fsS http://127.0.0.1:3000/health # zappier-edge (billing + users)
curl -fsS http://127.0.0.1:3010/health # account-balance (NATS SoT)
curl -fsS http://127.0.0.1:3011/health # customer-service
curl -fsS http://127.0.0.1:3012/health # sales-pricing
curl -fsS http://127.0.0.1:3100/health # middleware
```
Open:
- Customer portal (public door): http://127.0.0.1:3021/portal/ — sign up, copy `x-api-key`, **Statement** tab
- Loopback edge admin: http://127.0.0.1:3000/admin — **Statement** per customer (not a public door)
- Staff IAM: http://127.0.0.1:3028/
- CS review: http://127.0.0.1:3011/
- Sales review: http://127.0.0.1:3012/
- Accounting review/export: http://127.0.0.1:3013/
- API docs: http://127.0.0.1:3000/docs
Test env NATS is the **three LXC cluster** on `vmbr1`:
`nats://10.10.10.21:4222,nats://10.10.10.22:4222,nats://10.10.10.23:4222`
Host loopback `:4222` may still listen but modules no longer use it. Do **not** publish NATS on `vmbr0`.
## Operator console (monitor)
On the same machine (or SSH tunnel `ssh -L 3850:127.0.0.1:3850 user@host`):
```bash
cd "$MONOREPO/packages/verae-fleet"
node src/cli.js serve
```
http://0.0.0.0:3850/ — Fleet (green/yellow/red), Trace, Docs. Tree-node **min 3**; pause does not count. On NS1, `verae-keep` restarts workers if they crash unless you paused/stopped them here.
```bash
node src/cli.js status
node src/cli.js ssh-check ns1
```
## Point Zapier at the lab
1. TLS in front of `:3000` (Caddy).
2. Zapier app env: `MIDDLEWARE_BASE_URL=https://<zappier-host>` (edge meters, then proxies to middleware).
3. Connect with the portal API key.
## Production (pick one)
| Where | Read |
|-------|------|
| Docker | [02-docker.pdf](02-docker.pdf) |
| Proxmox | [03-proxmox.pdf](03-proxmox.pdf) |
| Cloud VMs | [04-virtual-servers.pdf](04-virtual-servers.pdf) |
| Dedicated / NS1 | [05-dedicated-hardware.pdf](05-dedicated-hardware.pdf) |
| Env vars & NATS | [06-linking-services.pdf](06-linking-services.pdf) |
| Backup / upgrade | [07-maintenance.pdf](07-maintenance.pdf) |
## Daily monitor
| Check | Expect |
|-------|--------|
| `curl -fsS https://zappier.example.com/health` | 200 |
| `curl -fsS https://middleware.example.com/health` | 200 |
| NATS `http://127.0.0.1:8222/healthz` on each node | ok (localhost only) |
| Fleet tree-node available ≥ 3 | green |
| Portal login + key still works | 200 `/v1/status` with `x-api-key` |
## If something is red
1. Fleet row color: yellow = paused/degraded, red = below floor — `restart` that replica.
2. 401/402 from Zapier → zappier-edge keys and rate card, not NATS.
3. Archive lookup empty → bloom miss is **silence**; check worm/tree floors, not “error replies”.