Bỏ qua nội dung
Lumen

Public status page

Nội dung này không tồn tại trong ngôn ngữ của bạn

The public status page is a single read-only route — https://<your-hub-url>/status — that anyone can visit without signing in. It shows the hosts you opt in plus their up/stale/down state and live CPU/RAM/disk. Use it for a small team’s “is the lab still alive” link or a public homelab status board.

Defaults are safe. New deployments do not publish anything. You have to (a) flip the global toggle and (b) tick at least one host before any name leaves the dashboard.

What the page shows

Per opted-in host, the page renders:

  • A state dot — up (fresh snapshot), stale (last snapshot older than 30s), down (the host is known but has nothing in memory), or unknown (newly created host, never seen).
  • The host’s name (the same string you typed in Settings → Hosts).
  • Last-seen timestamp in the visitor’s local timezone.
  • Three live numbers — CPU%, RAM%, Disk% — when state is up or stale.

Hosts in down or unknown state render with the dot + name + last seen only, so a visitor can tell “the host exists, it’s just not reporting right now” from “no host at this URL”.

The page polls /api/public/status every 15 seconds. There is no WebSocket — the page is meant to be left open for casual checking, not pushed to.

Configure

Sign in as admin, open Settings → Status page:

  1. Tick Publish the status page. You can save in either state; the rest of the form is required only when published.
  2. Set a Title (default: Status) and an optional Description (one short sentence is plenty).
  3. Tick Show on /status next to each host you want to expose. Hosts in this list aren’t filtered any further — every opted-in host is visible regardless of tags, rules, or silence state.
  4. Save. Visit https://<your-hub-url>/status to confirm.

The settings live in the same settings SQLite table as everything else; per-host opt-in is a public_visible column on hosts added in migration 0018.

What’s NOT shown

To avoid surprising leaks, the public page deliberately omits:

  • The host’s IP / hostname / OS / kernel / agent version (all those still appear on the admin-only host detail page).
  • Tags. The page never reveals a host’s tag set, so naming hosts after their role + tagging them with PII is safe.
  • Alert rules + recent alert events. The status page is “is the host alive right now”, not a partial alerts dashboard.
  • Container telemetry. Container names and resource usage are admin-only.
  • Charts. Only current values, never time series.
  • API keys, settings, the rest of the /api/* surface — those all require a session.

If you need any of the above on a public page, run a real status-page tool in front of Lumen and let it scrape /api/v1/hosts with a scoped API key. That keeps Lumen’s public surface narrow.

Hide / unpublish

Open Settings → Status page, untick Publish the status page, save. The page immediately returns {enabled: false} and the public URL renders a one-line “This status page isn’t published” notice. Per-host visibility stays saved so you can republish without re-ticking everything.

To fully clear the rows:

DELETE FROM settings WHERE key LIKE 'public_status.%';
UPDATE hosts SET public_visible = 0;

Caching, rate limit, abuse

The handler sets Cache-Control: no-store so a stale CDN edge can’t pin a sensitive snapshot longer than you intended. There is no per-IP rate limit yet — the expected audience is small. If you front Lumen with a reverse proxy that supports a rate limit (nginx limit_req, Caddy rate_limit, Cloudflare rules), apply it to /status and /api/public/status first; the rest of /api/* is already session/API-key gated.

Troubleshooting

/status shows “Status page is unreachable” : The frontend couldn’t reach /api/public/status at all. Check the hub is actually up and your reverse proxy isn’t blocking the path. Try curl https://<hub-url>/api/public/status directly.

/status shows “This status page isn’t published” : The global toggle in Settings → Status page is off, or the row was deleted from the settings table.

Host appears but state stays unknown even though it’s running : The agent isn’t actually reaching the hub — check docker logs lumen-agent-* for ingest failed, and verify the agent’s LUMEN_HUB_URL from inside the agent’s network. The status page reflects whatever the dashboard reflects.

State flips between up and stale : The agent’s collection interval is comparable to the 30s stale threshold. Tighten the interval (Settings → Runtime → Agent collection interval) or accept the flicker.