Deploy Runbook

If you don't have an account, sign up at cloudflare.com. Free tier is fine for v1.0 launch (we're well under all the rate limits).

If 613parts.ca isn't already in your sites list, you need to add it and switch nameservers at your registrar. This is the longest gate — DNS propagation can take 1–24 hours.

Cloudflare will scan existing DNS records and import them. Then it'll show you 2 nameservers (something like jonas.ns.cloudflare.com and kate.ns.cloudflare.com).

Log into wherever 613parts.ca is registered (GoDaddy, Namecheap, Domains.ca, etc.). Find DNS / nameserver settings. Replace whatever's there with the 2 Cloudflare nameservers.

Wait for propagation. Cloudflare emails you when active.

You need the repo cloned locally and a terminal open in the worker/ subdirectory.

cd /path/to/613parts/worker
node --version  # v20.x.x or higher
ls              # should show wrangler.toml, package.json, src/

From the worker/ directory:

npx wrangler login

What happens:

1. Wrangler prints a URL and waits
2. Your browser opens to a Cloudflare authorize page
3. Click Allow
4. Browser shows "Successfully logged in" → close it
5. Terminal prints Successfully logged in.

npx wrangler whoami

You are logged in with a Cloudflare API Token, associated with the email your@email.com.
👋 You are logged in with the email associated with the Cloudflare account: Your Account (acct-id)

Catalog data, AS400 invoice imports, computed driver routes, delivery events — all live in this bucket.

npx wrangler r2 bucket create 613parts-catalog

Creating bucket '613parts-catalog'...
Created bucket '613parts-catalog'.

KV caches geocoder results so we don't re-bill Google Maps for repeat customer addresses.

npx wrangler kv:namespace create SEARCH_CACHE

🌀 Creating namespace with title "613parts-bot-worker-SEARCH_CACHE"
✨ Success!
Add the following to your configuration file:
[[kv_namespaces]]
binding = "SEARCH_CACHE"
id = "a1b2c3d4e5f67890abcdef1234567890"

Copy the id value

Open worker/wrangler.toml and replace both occurrences of "REPLACE_WITH_KV_NAMESPACE_ID" with that id:

# Top-level (used by `wrangler dev`)
[[kv_namespaces]]
binding = "SEARCH_CACHE"
id = "a1b2c3d4e5f67890abcdef1234567890"

# Production (used by `wrangler deploy --env production`)
[[env.production.kv_namespaces]]
binding = "SEARCH_CACHE"
id = "a1b2c3d4e5f67890abcdef1234567890"

Two are required for the worker to deploy and run; the rest enable specific features.

Required — set these now

# Generate a strong secret first
openssl rand -hex 32  # copy the output

# Then set it
npx wrangler secret put JWT_SECRET --env production
# prompts: Enter a secret value: paste-the-hex-string

From dashboard.stripe.com → Developers → API keys. Start with the test key (sk_test_*). Swap for live (sk_live_*) when ready to take real money.

npx wrangler secret put STRIPE_SECRET --env production

Strongly recommended

# Generate, then set
openssl rand -hex 16
npx wrangler secret put CRON_TRIGGER_TOKEN --env production

In Slack: create a new app at api.slack.com/apps → Incoming Webhooks → activate → add to #ops-quotes channel → copy webhook URL.

npx wrangler secret put SLACK_WEBHOOK_URL --env production

From console.cloud.google.com → create project → enable Geocoding API → Credentials → create API key → restrict to Geocoding API only (good hygiene).

npx wrangler secret put GOOGLE_MAPS_API_KEY --env production

Optional — set when ready

npx wrangler secret list --env production

From worker/ directory:

npm install                           # first time only
npm run typecheck                     # should pass cleanly (0 errors)
npm run test:unit                     # 49 tests passing
npx wrangler deploy --env production

Total Upload: ~250 KiB / gzip: ~80 KiB
Uploaded 613parts-bot-worker (2.1 sec)
Published 613parts-bot-worker (3.4 sec)
  bot.613parts.ca (custom domain)
Current Deployment ID: a1b2c3d4-...

curl https://613parts-bot-worker.your-account.workers.dev/health

The worker is up at the workers.dev URL. Now route the production custom domain to it.

OR via the worker's settings:

Cloudflare auto-creates the DNS record + provisions an SSL cert. Takes ~30 sec.

curl https://bot.613parts.ca/health

One project per app. From repo root:

cd ../driver-app
npx wrangler pages project create 613parts-driver-app

cd ../dispatch-app
npx wrangler pages project create 613parts-dispatch-app

Each Pages project gets its own custom domain. Same flow as worker (Step 6) but on Pages.

Enter: drive.613parts.ca

Enter: dispatch.613parts.ca

Cloudflare provisions DNS + SSL in ~60 seconds for each.

This is a separate clicking-through-ManyChat task. Worker URLs are already production-correct (bot.613parts.ca) so the External Request bodies in the guide work as-is.

Open the flow guide:

That doc has 12 sub-steps: 25 Custom Fields, 4 keyword triggers, 5 flows, postback dispatcher, smart filter for live-chat. Plan ~90 min uninterrupted.

Stream live logs from the worker

npx wrangler tail --env production

Useful when something breaks in production and you need to see what's happening in real-time.

Roll back the worker to a previous deploy

npx wrangler deployments list --env production
# Pick a previous deployment ID, then:
npx wrangler rollback [deployment-id] --env production

Roll back a Pages app

Manually trigger the catalog/invoice cron

curl -X POST https://bot.613parts.ca/cron/run \
  -H "authorization: Bearer $CRON_TRIGGER_TOKEN"

Used when AS400 missed the nightly run and you need today's routes processed manually.

Force a worker redeploy without code changes

cd worker
npx wrangler deploy --env production --compatibility-date=$(date +%Y-%m-%d)

Update a single secret

npx wrangler secret put SECRET_NAME --env production
# Overwrites whatever was there. No downtime.