Asclepius

The one that asks

Asclepius is the only part of the platform that goes and checks. Everything else answers when asked; this is the thing doing the asking. It polls each application, keeps the history, and raises an alert when something stops responding, so the first person to notice an outage is not a customer.

What it gives you

Everything watched in one list. Platform applications appear automatically, because they are already registered elsewhere and this mirrors them in. A service that is not part of the platform, a client's own system for example, is added by hand and is then treated exactly the same.

State, and how long it has been that way. Each row says whether the service is up, down or deliberately paused, and since when. "Down for two minutes" and "down since Tuesday" are very different problems and the list distinguishes them at a glance.

Checks you can pause. A service being worked on can be taken off the watch without being removed from the list, so maintenance does not generate noise and nobody has to remember to add it back afterwards.

A dry run before you commit. An address can be tried once, from the host, to see what the watcher would record, which turns "why does it think this is down?" into something you can answer in a few seconds.

Alerts through the platform's own channels. When something stops answering, the notification goes through the same delivery system as everything else, with the same guarantee that a message nobody could deliver is not a message quietly lost.

Who it is for

Anyone running services they would rather hear about from their own tooling than from a customer. It is also the piece that makes watching a client's systems a normal part of the service you offer them, rather than a separate product you have to buy.