Documentation
How Perstat works, in the same register the product uses: concrete numbers, real limits, and honest gaps. Everything here describes what ships today.
These docs cover the product as it is, not as it is planned. Where a capability does not exist yet, the page says so and the changelog carries it when it lands. Configuration happens in the web app today; a public API with scoped keys is the stated next step.
- Getting started
From signup to a running check in a few minutes, and what a sensible first hour looks like.
- Monitors and check types
12 check types, what each verifies, and the knobs that matter: assertions, inline subchecks, dual-stack behavior, intervals.
- Quorum and regions
The rule that decides when an incident exists: regions vote, a quorum confirms, and everything below quorum stays visible without paging anyone.
- Alerting and on-call
Escalation that walks push, SMS, phone call until someone acknowledges, and the semantics that keep a loud night from becoming a chaotic one.
- Incidents and post-mortems
Confirmed incidents open themselves; manual incidents, post-mortems, and action items keep the story in one place instead of scattered across chat.
- Status pages
Public or gated pages on your own domain, with subscribers who opted in twice, version history you can restore, and a record that cannot drift from your monitoring.
- Maintenance and curation
Planned work does not count against you, false positives can be discarded with a reason, and every correction is logged and reversible.
- Evidence and retention
Exclusion windows with real boundaries, availability floored instead of rounded, and retention that is enforced twice.
- Host agent
One line to install, signed root-free self-updates, and a strict data-minimalism rule: the agent never reads command lines, environment variables, or usernames.
- Plans and limits
The hard numbers per plan, the same ones the product enforces. If a table here and the product ever disagree, that is a bug and we say so.