Gatus

Gatus is a health dashboard aimed at developers, and its README opens with the objection you would raise anyway. Why not use Alertmanager, CloudWatch or Splunk. The answer is that those rely on existing traffic, so if nobody is calling a broken endpoint, nothing fires and your clients find the outage before you do.

Gatus calls the endpoints itself on a schedule using HTTP, ICMP, TCP or DNS queries, then evaluates the result against conditions you write. A condition can test the status code, the response time, the body, the certificate expiration or the resolved IP address, which makes the same mechanism usable for automated user acceptance tests.

The design choices show up in daily use.

• Readable configuration: adding an endpoint to monitor is a short block of YAML, not a project.
• Alerting breadth: Slack, Teams, PagerDuty, Discord, Twilio, Telegram, Opsgenie and dozens more, plus custom providers.
• Low resource use: a Go binary small enough to sit in a cluster without anyone noticing it.
• Badges: uptime and response time badges served from the API for embedding elsewhere.
• Security options: basic authentication, OIDC, TLS encryption and exported metrics are all configurable.

Storage, maintenance windows, announcements, client behavior, tunneling and custom labels are configurable too, alongside experimental support for remote instances and an alpha suites feature for grouping related checks.

A single docker run command starts it on port 8080, which is enough to have real endpoints monitored within a minute. Its author runs it in a Kubernetes cluster watching his own core applications.