Incident Management and Data Downtime Tracking

Data incidents (quality issues, downtime, pipeline failures) are not being properly tracked or communicated in DataHub, or the incident workflow is not clear. Organizations need a systematic way to document and resolve data issues.

You Might Be Asking:

• How do I create and track data incidents?
• Can DataHub integrate with our incident management system?
• How do I notify stakeholders about data issues?

1. Create incident via UI:

1. Navigate to the affected dataset
2. Click "Incidents" tab
3. Click "Raise Incident"
4. Fill in details:
   - Title and description
   - Severity (P0-P4)
   - Type (Operational, Quality, Schema Change, etc.)
   - Affected entities
5. Assign to owner/team
6. Add relevant stakeholders
7. Save and notify

2. Create incident via GraphQL:

mutation raiseIncident {
  raiseIncident(
    input: {
      type: DATA_QUALITY
      title: "Sales table contains negative amounts"
      description: "Discovered negative values in amount column"
      priority: 1  # P0 = 0, P1 = 1, etc.
      resourceUrn: "urn:li:dataset:(urn:li:dataPlatform:snowflake,db.schema.sales,PROD)"
      status: ACTIVE
    }
  ) {
    urn
  }
}

3. Automate incident creation from assertion failures:

# DataHub Actions configuration
action:
  type: "raise_incident"
  config:
    # Trigger on assertion failure
    filters:
      - type: "assertion_result"
        status: "FAILURE"

    # Auto-create incidents for critical assertions
    severity_mapping:
      CRITICAL: 0  # P0
      HIGH: 1      # P1
      MEDIUM: 2    # P2

    # Notification
    notify_owners: true
    slack_webhook: "${SLACK_WEBHOOK}"

4. Update incident status:

mutation updateIncident {
  updateIncident(
    input: {
      urn: "urn:li:incident:12345"
      status: RESOLVED
      message: "Fixed by correcting upstream ETL job"
    }
  )
}

5. Integrate with external incident management:

# Listen to DataHub incident events and forward to PagerDuty/ServiceNow
from datahub.emitter.kafka_emitter import DatahubKafkaEmitter

def on_incident_created(event):
    # Forward to PagerDuty
    pagerduty_client.create_incident(
        title=event['title'],
        service_id=SERVICE_ID,
        urgency='high' if event['priority'] <= 1 else 'low',
        body=event['description']
    )

# Subscribe to incident events
# Subscribe to MetadataChangeLog_v1 topic, filter for incident entities

6. Query incident history:

query {
  dataset(urn: "YOUR_URN") {
    incidents(start: 0, count: 20) {
      incidents {
        urn
        type
        title
        status
        created {
          time
          actor
        }
        resolved {
          time
          actor
        }
      }
    }
  }
}

Use incidents to track data quality issues, pipeline failures, schema breaking changes, or any data-related problems. Link incidents to affected datasets for full context. Consider integrating with your existing incident management workflow.

• Incidents Guide
• Observability