Skip to main content

Incident Creation

Incidents in Aurora are automatically created when alerts are received from integrated monitoring platforms. There is no direct POST endpoint for manual incident creation.

How Incidents are Created

Incidents are created automatically through the following flow:
  1. Alert Reception: Aurora receives an alert from an integrated platform (Grafana, Datadog, PagerDuty, etc.)
  2. Correlation Check: The system checks if the alert should be correlated with an existing incident
  3. Incident Creation: If no correlation is found, a new incident is created
  4. RCA Initiation: Aurora automatically begins Root Cause Analysis (RCA)

Supported Alert Sources

Aurora creates incidents from the following monitoring platforms:
  • Grafana - POST /api/grafana/alerts
  • Datadog - POST /api/datadog/webhook
  • Netdata - POST /api/netdata/alerts
  • PagerDuty - POST /api/pagerduty/webhook
  • Splunk - POST /api/splunk/alerts
  • Jenkins/CloudBees - POST /api/jenkins/deployment or POST /api/cloudbees/deployment
  • Dynatrace - POST /api/dynatrace/problems
  • BigPanda - POST /api/bigpanda/webhook

Incident Schema

When an incident is created, it includes the following fields:
UUID
Automatically generated unique identifier
string
User who owns this incident (from the alert webhook authentication)
string
The monitoring platform that generated the alert
integer
ID of the alert in the source monitoring system’s database
string
default:"investigating"
Initial status is always investigating
string
Severity level extracted from the source alert
string
Title/summary of the alert
string
Service or component affected
string
Environment where the alert occurred (production, staging, etc.)
string
default:"idle"
Status of Aurora’s RCA process. Initial value is idle, then transitions to running
timestamp
When the incident started (from alert timestamp)
string
default:"thoughts"
Default UI tab is thoughts

Database Table Structure

Incidents are stored in the incidents table with the following schema:

Automatic RCA Process

Once an incident is created, Aurora automatically:
  1. Creates a chat session for the RCA investigation
  2. Analyzes the alert payload and context
  3. Generates investigation thoughts and suggestions
  4. Executes diagnostic commands (if configured)
  5. Provides a summary and potential remediation steps

Correlation and Merging

If Aurora detects that a new alert is related to an existing incident:
  • The alert is added to the incident_alerts table linked to the existing incident
  • The existing incident’s correlated_alert_count is incremented
  • The alert’s service is added to affected_services array
  • No new incident is created

Manual Incident Merging

You can manually merge incidents using the merge endpoint:
Incidents can be merged via the Update Incident endpoint by specifying alert correlation.

Integration Setup

To enable automatic incident creation, configure webhooks in your monitoring platforms to point to Aurora’s webhook endpoints. See the Integrations Overview documentation for platform-specific setup instructions.