Rules for the way you watch
Service-Level, Error Rate, Latency, Log Pattern, Infrastructure, Workspace-wide, Environment-Specific and Saved Query. Set an evaluation window from 1 to 60 minutes.
Platform
Know when a service breaks, and keep the response in one place.
Alert rules check your telemetry every minute. When a condition is met, the rule opens an incident for that service and notifies the channels you ticked. When the condition clears, the incident resolves itself.
What you get
Tell people where they already look
Send to a webhook, email, Slack or PagerDuty, up to 20 channels per rule. Each one gets a message when the rule fires and another when it recovers.
Run the response from one page
An incident moves from Open to Acknowledged, Investigating, Monitoring and Resolved. Admins can also declare one by hand, with a severity from SEV1 to SEV4, a commander and up to 20 services.
How it works
Create a rule
Choose what to watch, how long a window to judge it over, and which channels to notify.
It opens one incident per service
When a rule goes from OK to firing, it opens an incident for the affected service.
See what was happening
The incident lists logs, traces and findings for the first affected service, from five minutes before to an hour after.
It closes itself
When the condition clears, the rule resolves the incident and marks it Recovered.
Limits worth knowing
- Incident changes do not notify. Declaring, acknowledging or resolving an incident does not send a message. Rules do.
- Public HTTPS only. Webhook and Slack destinations must be public https addresses on port 443.
- Rules are not edited in place. After you create a rule you can change its channels, pause it or remove it. Other changes mean a new rule.
- Not enough traffic is not OK. A rule with too few requests says so, rather than reporting OK.
Get paged for the right things.
Create a rule in a trial workspace and point it at Slack.