Alerts and notifications

Incidents

An incident opens when a rule fires and closes on its own when the condition clears. You do not acknowledge or dismiss them — a machine that came back does not need you to confirm it came back.

The date filter hides open incidents

Incidents are filtered by when they opened. An incident that started before the window you are looking at and is still open does not appear — and those are usually the ones that matter most.

If a machine has been down for a week and you are looking at “last 24 hours”, you will not see it. Widen the window.

What arrives, and where

Notification channels are configured per account, not per machine. Telegram and email both work.

What gets sent:

• A rental starts — the rate and an estimate of what it will pay.
• A rental ends — how long it ran and what it earned.
• A machine goes offline — after the 300-second threshold, not on the first missed heartbeat.
• Plan and billing changes — including the grace-period deadline before an account freezes.

Telegram

Connect once from the panel; the bot then messages you directly. The link is to your Telegram account rather than to a specific chat, so it survives you changing devices.

Choosing what actually reaches you

The failure mode of notifications is not missing one — it is getting so many that you stop reading them. Two habits help:

Let offline alerts be loud and rentals be quiet. A machine that fell over needs you now; a rental that started does not.

Fix the noisy machine rather than muting the alert. A machine that flaps offline twice a day is telling you something real about its network or its power.

A machine is offlineWhat to check, in the order that finds it fastest.