UptimeGuard
Uptime monitoring service that pings endpoints on a schedule and alerts the moment something goes down.
the problem
Finding out an endpoint is down usually comes from a customer complaint, not a monitor -- cheap uptime tools tend to be either too basic (a single current-status flag) or too expensive for a side project.
approach
A Bun + Elysia backend pings every monitored URL on a schedule with a 5-second timeout, then persists each check -- up/down, HTTP status code, latency -- as its own row via Prisma against Postgres, so history is queryable, not just a rolling "currently up" flag.
architecture
Elysia on Bun
A lightweight, TypeScript-native HTTP framework on the Bun runtime instead of a heavier Node/Express stack.
Per-check persistence
Every check (up/down, status code, latency) is its own Prisma row, so real history and trends are queryable, not just a live flag.
Timeout-bounded checks
A 5-second AbortSignal.timeout on every fetch, so one slow or hanging endpoint can't stall the whole check loop.