The investigation starts before anyone opens a laptop.
An alert from your Alertmanager, or an e-mail from a tool that cannot call a webhook, starts the investigation and posts a short summary to your team’s channel.
Rules you write
Match alerts by name, severity, namespace or label. A rule investigates on its own or waits for someone to press a button.
Triage before the model
A quick read of the cluster gives each alert a priority. A rule can investigate only from a priority upwards.
A storm stays one job
A cooldown per alert group and a daily cap keep an alert storm from turning into a hundred investigations.
Into the channel, then to people
Cards and summaries go to Slack or Google Chat. An escalation channel hands the alert to people with what SKYLINE found.
Set it up in three steps
- 01
Give Alertmanager a webhook URL
Each environment has its own URL. Add it as an extra receiver and keep the ones you have.
- 02
Add a rule
Start from the template for paging alerts: severity critical or page.
- 03
Read the result
In the console and in the channel. Every alert group also opens an incident.