Create a monitor

Monitors are the building block of Spectra: every check, alert, and status-page component starts from one. This page covers every setting on the create-monitor screen. If you just want the fast path, take the Quickstart.

To start, open Monitors → Add monitor in the app.

1. Choose what to monitor

First, pick a monitor type. Each type checks a different kind of failure, so start with the signal that best describes “down” for your service:

Type	Use it when	Checks
Website	A public website or API endpoint	HTTP/HTTPS status, response time, keyword
API (paid)	An endpoint that must return the right data	Method, auth, request body, JSON assertions
Ping	A host must be reachable	ICMP reachability, latency, packet loss
Port	A TCP service must accept connections	TCP connect on a given port
Heartbeat / cron	A scheduled job must run	Inbound ping within a schedule + grace period
SSL & domain	A certificate or domain must not expire	Certificate + domain expiry, chain, hostname

2. Name the monitor

Give it a human-readable name like API or Marketing site. It appears in your dashboard, alerts, and reports, and can be renamed later without affecting its history.

3. Set the target

• Website / API — the full URL, starting with http:// or https://.
• Ping — an IPv4/IPv6 address or hostname (re-resolved via DNS each check).
• Port — a host plus a port (enter a custom port or pick a preset like PostgreSQL 5432 or Redis 6379).
• Heartbeat — no target; Spectra gives you a unique ping URL.
• SSL & domain — the domain to inspect.

4. Choose regions

Regions are the probe locations your checks run from. Checks rotate through your selected regions, and Spectra confirms DOWN from 2 or more regions before alerting — which prevents false positives from a regional network blip.

• The Free plan checks from a fixed region; paid plans let you choose from the global list (North America, Europe, Asia-Pacific, and more).
• Pick regions close to your users — response time is measured from the checking region, so a distant probe inflates your latency numbers.

5. Pick a check interval

How often the monitor runs: 30 seconds · 1 minute · 5 minutes · 15 minutes · 30 minutes. The Free plan’s minimum is 5 minutes; faster intervals are available on paid plans. Tighter intervals catch short outages but use more of your check volume.

6. Set a timeout

How long to wait for a response before the check counts as failed: 10, 30 (recommended), or 60 seconds. Set it above your service’s normal worst-case response time so a slow-but-working response isn’t treated as down.

HTTP advanced settings (Website & API)

• Redirects — choose whether the monitor follows HTTP redirects (e.g. a 302) or treats them as the final response.
• Expected status code — by default any 2xx counts as up. For an auth-protected route that should return 401, set that as the expected code so the monitor is up only when it answers exactly as intended.
• Keyword / text assertion — require text that must (or must not) appear in the response body. This catches error pages served with a 200 and empty renders — the monitor goes DOWN even if the status code is OK.
• Custom request headers — add headers such as Authorization: Bearer … for authenticated endpoints. Credentials are stored encrypted and never shown in alerts.
• API only — choose the method (GET/POST/PUT/…), add a request body, and assert on the JSON response using dot notation (e.g. data.user.active == true).

7. Configure alerts

Decide when and where you’re notified:

• Alert delay — notify after the failure persists (for example after 2, 5, or 10 minutes), and re-notify on an interval (e.g. every 30 minutes) until it recovers.
• Channels — attach the alert channels this monitor should notify (Slack and email today). A channel is reusable across monitors.

8. Save

Click Save. The first check runs immediately, metrics start computing, and you land on the monitor’s detail view with its graphs, logs, and incident history.

Troubleshooting

• Monitor shows DOWN but the site works in my browser — check for a keyword/status-code assertion that isn’t matching, a firewall blocking Spectra’s probe IPs, or a slow response exceeding your timeout.
• Response times look higher than expected — you may be checking from a region far from where the service is hosted. Add a closer region.
• No alert fired during an outage — confirm a channel is attached to the monitor and that quiet hours or severity filters aren’t suppressing it.

Next steps

• Dig into a specific type: Website · API · Heartbeat · SSL.
• Set up Slack and tune quiet hours & severity.