skytrace
Explore blocks, labels, lists, and posts on Bluesky. skytrace.aly.town
typescript react bluesky atproto
Star
Fork
Watch
skytrace docs deployment.md
2.2 kB · 50 lines
Markdown
at main

Deploy SkyTrace #

Run the container #

docker run -p 8080:80 ghcr.io/alyraffauf/skytrace:latest

Open http://localhost:8080. Pass environment variables with -e to change settings. See the configuration guide for the full list. Change requires recreating the container.

Publish a static build #

bun install --frozen-lockfile
bun run build

Copy the dist/ folder to your static host. If you need custom settings, edit public/runtime-config.js before running bun run build, or edit the deployed runtime-config.js directly.

Hosting setup #

Your host needs three things:

  1. Route all app paths to index.html so the router can handle them.
  2. Serve fingerprinted assets with long cache headers.
  3. Return 404 for unknown routes, not a catch-all redirect.

Two files in the repo handle this for you:

  • public/_redirects — rewrites app routes to index.html. Use this on Netlify or Cloudflare Pages.
  • public/_headers — sets Cache-Control and X-Robots-Tag: noindex for profile and list URLs.
  • nginx.conf — the container's equivalent rules, including a one-year cache for fingerprinted assets.

If your host does not support _redirects or _headers, translate the rules into its config. The key point is that unknown routes must return a 404 with 404.html. A catch-all rewrite to index.html breaks things.

Whenever you add a new app route, update _redirects, nginx.conf, and _headers if the route needs different headers.

Verify the deployment #

Hit these URLs after deploying:

  • / returns 200 with no noindex header.
  • A profile tab URL and a list URL return 200 with X-Robots-Tag: noindex.
  • An unknown route, a missing asset, and /sitemap.xml return 404.
  • /runtime-config.js returns JavaScript with Cache-Control: no-store.
  • /robots.txt returns plain text and allows crawling so the noindex headers on profile URLs are read correctly.

If Cloudflare adds a managed crawler policy to robots.txt, leave it alone. Use Search Console's URL Inspection to confirm Google can render the homepage.

Vite's dev server does not apply these rules.