Uptime checks & status pages

Alongside the agent (inside-out), the hub can also check outside-in that your services are reachable: HTTP(S) and TCP checks that run from the hub, with no agent on the target server. The results feed a public status page for customers or colleagues.

Creating checks

1. Sidebar → Uptime
2. Click + Check and fill in:
   • Name: free text, also shown on the status page
   • Type: http (follows redirects, validates TLS) or tcp (port open?)
   • Target: full URL (https://…) or host:port for tcp
   • Interval: 30 to 3600 seconds
   • Timeout: 1 to 60 seconds, always shorter than the interval
   • Expected status (http): for example 200

Limits: at most 50 checks per tenant. Targets that point at internal or private IP ranges (loopback, RFC1918, link-local) are rejected: the hub will not act as an SSRF proxy.

How the hub decides something is down

• The worker schedules every check on its own interval (scheduler tick: 15 s).
• A check only flips to down after 2 consecutive failures. A single network hiccup does not raise an alert (flap protection).
• On down, a critical alert (uptime.down) enters the regular alert pipeline: same notification channels, same acknowledge flow.
• When the check recovers, the alert resolves automatically.

TLS expiry detection

On every https check the hub reads the certificate chain. If the certificate expires within 14 days, a warning alert (uptime.cert_expiring) appears with the expiry date. After renewal the alert clears itself on the next check.

Public status pages

1. Sidebar → Uptime → Status pages section
2. Pick a title, a slug and the checks to show.
3. The page is immediately publicly reachable at:

https://app.monsys.ai/<locale>/s/<slug>

Machine-readable (JSON, for your own integrations):

https://api.monsys.ai/v/status/<slug>

What visitors see: each check’s name, current status and 30-day uptime percentage. What they never see: the target itself (URL, hostname, port). Status pages leak no infrastructure.

Status page visibility

Per page you choose how it can be reached:

Mode	Behaviour
public (default)	anyone with the slug sees the page
token	only via a secret link …/s/<slug>?token=<token>; the token can be rotated with Rotate token — the old link stops working immediately
password	visitors enter a password (8–128 characters); the hub stores only a bcrypt hash and verifies server-side

Switching from public to token or password makes the public URL (and the JSON endpoint) return 404/401 right away for anyone without the key.

Roles

Role	Can
Viewer	view checks and results
Editor	create, edit, delete checks and status pages
Admin	same

Relation to agent monitoring

Aspect	Agent (inside-out)	Uptime check (outside-in)
Requires agent on the host	yes	no
Sees	CPU, disk, services, logs, CVEs	reachability and TLS as a visitor sees them
Detects	the cause	the symptom

Use both: the uptime check tells you a service is down, the agent tells you why.