Definition of done

Incidents can be programatically resolved using done definition. When an evidence is added to an incident, a done definition can be created based on the attached object of the evidence.

Done definitions are checked periodically every 5 minutes.

Examples:

A. Using config analysis

The following expression for a done definition using a config analysis, will mark the incident as resolved if the linked config analysis is resolved.

analysis.status == 'resolved'

B. using config item

Assume an incident where a postgres database fails to accept more than 20 concurrent connections. We can setup a done definition on the related postgres config as follows

config.max_connections > 30

Now, whenever the postgres config is modified to have more than 30 max connections, the incident will be marked as resolved automatically

C. Using health check

Assume an incident where the reverse proxy is returning too many 5XX errors. An evidence is provided linking a health check for the backend API service. We can setup a done definition as follows

check.status == 'healthy'

# OR

check.status == "healthy" && check.age > duration("30s")