Zeish Status

Uptime status page for Zeish infrastructure. No servers, no third-party monitoring SaaS — just GitHub Actions checking each endpoint and committing the results, and GitHub Pages serving the generated page.

How it works

• config/services.json lists the monitored endpoints, each tagged with a product (zeish / arin) and kind (frontend / backend). The page groups cards by product and has client-side filter pills to show just one product at a time.
• scripts/healthcheck.ts pings each one, appends the result to data/<slug>/pings.json (rolling ~2 day window) and data/<slug>/daily.json (90-day uptime rollup), and opens/closes entries in data/incidents.json when a service's status flips.
• scripts/build.ts renders data/ into a static docs/index.html, styled to match zeish.dev's dark theme.
• .github/workflows/healthcheck.yml runs on GitHub's shortest cron interval (5 minutes) but loops 5 times internally with a 60s sleep, committing after each pass — so checks land roughly once a minute even though the scheduler can't trigger that often.

Scripts are plain .ts files run directly with node — Node 24 strips TypeScript types natively (no ts-node/tsx, no build step, no extra dependencies).

Everything (data + the built page) lives in git history on main, so GitHub Pages just needs to serve /docs — no separate deploy step.

One-time setup after creating the GitHub repo

Settings → Pages → Source: Deploy from a branch → Branch: main, folder: /docs.

Custom domain: status.dvito.cloud. scripts/build.ts writes docs/CNAME on every build (it has to — docs/ is regenerated from scratch each run, so the CNAME file would otherwise get wiped the first time the workflow commits). In Cloudflare DNS, add:

Type	Name	Target	Proxy status
CNAME	status	spinupdev.github.io	DNS only (grey cloud)

Keep it DNS-only until GitHub finishes issuing the HTTPS certificate for the domain (Settings → Pages will show "DNS check successful" then an "Enforce HTTPS" checkbox) — Cloudflare's proxy can interfere with that initial validation. Safe to switch to proxied afterwards if you want.

Local usage

npm run check   # run one round of health checks against config/services.json
npm run build   # regenerate docs/index.html from the current data/

Notes

• A check counts as "up" if the host returns any status code under 500 (i.e. it's reachable and answering), not necessarily HTTP 200 — endpoints without a root route would otherwise show as permanently down.
• Repo must stay public for unlimited free GitHub Actions minutes; a private repo would burn through the 2,000 free minutes/month in a few days at this check frequency.
• Zeish Backend (api-edge.dvito.cloud) and Arin Frontend (arin.dvito.cloud) share the same origin host, so they tend to go down together — that's expected, not a bug in the checker.