Flapping never wakes you up
A failed check is retried, an outage needs several failures in a row and, if you run peers, a second place to agree. Ten services down behind one database send one alert: the cause.
Self-hosted uptime monitor
One small binary checks your services, keeps their history, and serves a calm status page. It wakes you only when something has really broken, never for a blip.
# one container, about 15 MBdocker run -d -p 8787:8787 \ -v "$PWD/hora-config:/etc/hora" \ -v hora-data:/data ghcr.io/uplg/horaEverything is running.
Flapping never wakes you up
A failed check is retried, an outage needs several failures in a row and, if you run peers, a second place to agree. Ten services down behind one database send one alert: the cause.
One small binary
HTTP, TCP, ping, DNS, push and exec checks, SQLite, the status page and ten alert channels in one static binary. The Docker image is about 15 MB. No Node, no Redis, no browser in the background.
Targets you can keep
Set 99.9 % and Hora shows the downtime you have left, warns when it burns too fast, and writes the monthly report for you.
A page people trust
Server-rendered, no JavaScript framework, readable on a phone. Also as plain text for curl, an Atom feed, JSON and badges.
The whole path, with the settings behind each step, is in Concepts.


What visitors see, what you see once signed in, and the plain-text view are described in The status page.
Hora is for people who want few alerts, each one worth getting up for, and their monitoring in a file next to the rest of their infrastructure. If you would rather click, or need a check type Hora leaves out, it has good neighbours: When Hora fits says which, and when.
Named after the Horai, the Greek goddesses of the hours. Its owl dozes while all is well.