Cloudflare error 1033 / HTTP 530 on a tunneled site: the cloudflared connector is gone, not DNS
Symptom: domain resolves to Cloudflare IPs, curl returns HTTP 530 and the body says error code: 1033. That means the named Cloudflare Tunnel has zero active connectors; DNS and the origin config are fine.
Diagnosis: cloudflared tunnel list shows the tunnel with an empty CONNECTIONS column. Check the ingress in the tunnel's config yml (e.g. ~/.cloudflared/<name>.yml) to see which local port it targets, then lsof -nP -iTCP -sTCP:LISTEN to confirm whether that origin process is also dead. In my case both the connector and a vite preview origin had died with a closed terminal while the API backend survived.
Fix: restart origin + connector detached, e.g. nohup npx vite preview --host 127.0.0.1 --port 4173 --strictPort & and nohup cloudflared tunnel --no-autoupdate --config ~/.cloudflared/<name>.yml run &. Look for 4x Registered tunnel connection in the log; the site returns 200 within seconds.
Gotchas: vite preview reuses server.proxy, so an env-driven proxy target (e.g. /api -> backend port) must be set when launching preview too. With ingress hostname routing to localhost vite, set originRequest.httpHostHeader: localhost or vite's host check may reject the public hostname. For durability, run cloudflared as a service (cloudflared service install) or host the origin somewhere that is not a laptop.