Choose who's told

Where alerts go

Each project sets this in Settings, under Who gets told, with the admin role. Each row there shows what that route answered last time.

E-mail, Telegram and Slack are the alert channels. They tell a down when the project has no Incidents, or when Incidents doesn't take it. They also carry what isn't an incident: reminders, a certificate about to expire, a heartbeat never pinged, and a word when Incidents refuses Status's key.

A down counts as told once a channel took it. If none did, Status tries again after 1, 2 and 5 minutes, then every 10 minutes.

Set up Telegram

1. Make a bot with Telegram's @BotFather and copy the token it gives you.
2. Add the bot to the group or channel the alerts should go to.
3. In Status, open Settings, choose Set up on the Telegram row, and enter the Chat id and the Bot token.

The chat id is a number (a group's starts with -100) or a public channel's @name. One way to find a group's: send a message in it, then open this address with your bot's token, and read chat.id:

https://api.telegram.org/bot<your bot token>/getUpdates

The token is kept encrypted and never shown again. Leave the field blank when you change the chat, and the token stays.

Set up Slack

1. In Slack, make an incoming webhook for the channel the alerts should go to. Its address starts with https://hooks.slack.com/services/.
2. In Status, open Settings, choose Set up on the Slack row, and paste it as the Incoming webhook.

The webhook address is kept encrypted, and only its last four characters are shown afterwards.

Connect Incidents

When the project has Incidents too, connect it on the Incidents row of Settings:

1. In Incidents, open the same project's Settings → Keys and make a server secret.
2. In Status, choose Connect on the Incidents row. The address is suggested; paste the secret and save. Status asks Incidents first and saves it only once Incidents takes it. Nothing is opened there.

With Incidents connected:

• a down opens an incident named after the monitor ("API is down", or "API is down in staging"), with the monitor's severity, and its recovery resolves it,
• each downtime links to what else happened then, in Incidents,
• the incidents a responder marks "Show on the status page" there appear on your status page.

If Incidents later refuses the secret, Settings says so, the alert channels hear it once, and the downs go to them instead.

Send a test alert

Send a test alert is in Settings, next to Who gets told, and on the setup guide. It sends "Test: your project's alerts reach you" on every alert channel, and asks Incidents whether it takes the secret, opening nothing. The message after it says which channels took it, and why any didn't.

Each webhook endpoint has its own Send a test: see Add an endpoint.

When people are told

A down is told within a minute of the check that confirms it. What a monitor and its project change about that:

• Tell when it's been down for: wait 1 minute, 3 minutes, 5 minutes, 10 minutes, 15 minutes or 30 minutes first. A downtime shorter than the wait counts in the uptime and wakes nobody.
• Depends on: another monitor of the project. While that one is down, this one's down isn't told: the other's alert is the news.
• Severity: sev1, sev2 or sev3. It doesn't change who is told; it's the incident's severity in Incidents, and is sent in webhooks.
• Environment: named in every alert and incident outside production.
• Maintenance windows: a down inside one is told to nobody. If it's still down when the window ends, it's told then. See Plan maintenance.
• Many at once: 3 or more of a project's monitors going down together are told in one alert.

A recovery goes the same way as its down. A down that came back before anyone could be told is told by its recovery only, which says both.

Reminders and I'm on it

While a monitor is down, its page has I'm on it, for anyone who may open the project. Until someone presses it, the alert channels that told the down hear it again after 30 minutes, then every 2 hours. Once someone does, the reminders stop, the channels hear who it is, and the monitor's pages say so.

A monitor that keeps flapping

A monitor going down for the 4th time within an hour is flapping. Status sends one alert saying so, then stays quiet until the monitor holds, up or down, for 15 minutes, and tells that: "Settled" or "Still down".

The weekly digest

Every Monday at 07:00 Helsinki time, everyone who may open the project gets last week by email: the project's uptime and each monitor's, the worst downtimes, the slowest answers, what is still down, and the certificates that expire within a month.