The load balancer reports healthy, users hit errors, and the team finds out from a ticket. Meanwhile a DNSSEC signature lapses or a DMARC record is edited, and nothing fails a build or pages anyone.
A health check behind the load balancer answers for the process, not for the path your users take through DNS, TLS, and the edge. The two disagree exactly when it matters.
Posture changes do not look like outages. A DNS record edited by hand, an SPF record that grew past the lookup limit, a DMARC policy quietly relaxed: none of these break anything today, and each one is found later by someone else.
During an incident, people need somewhere to look. Without a page that says what is known, the answer is the support queue.
Dependably StatusCheck probes the assets you operate from outside your network, the way your users reach them, and treats a change in state, grade, or record as the event to tell you about.
Uptime probes run over HTTP, TCP, or ICMP, every 60 seconds by default. A failure is retried before anything is declared down, so a single dropped packet is not an incident.
HTTP · TCP · ICMPDNS records and DNSSEC are checked on their own cadence, and SPF, DKIM, and DMARC are read for each domain. Each endpoint's TLS configuration carries a grade from A+ to F. A grade change, a changed record set, or a failing DNSSEC signature is an event.
DNSSEC · SPF/DKIM/DMARCAn alert fires when a state changes, and a failed send is retried rather than lost. Alerts go to Slack, PagerDuty, Microsoft Teams, Discord, Telegram, or a webhook; status page subscribers are notified by email.
alertThe same change updates a public status page that ships without scripts. Schedule maintenance windows ahead of time and the page shows them while they run, post written updates as an incident unfolds, and subscribers hear when a host goes down or recovers. Maintenance windows are excluded from the uptime figure at query time, so the number can always be recomputed.
status pageEvery probe result within the retention window you set is kept, so uptime for any period in it is a query over the record, with maintenance windows excluded when you ask rather than baked in.
Dependably StatusCheck is self-hosted and open source under Apache-2.0, and probes from wherever you run it. Our own public status page runs on it, so you can see it working before you install anything, and you can get in touch about your deployment.