Set up alerting
Set alerting up last, once there is somewhere for its alerts to go. AWS has native discovery and alert setup in Shankh; Grafana, Jenkins and SigNoz are connected as Vörr connectors; anything else sends to the private webhook.
AdminCloud Integration, including alert-management credentials, requires Admin access.
Check where your source belongs
| Setup path | Sources |
|---|---|
| Native Shankh setup | AWS |
| Vörr connector | Grafana, Jenkins, SigNoz |
| Private webhook | Other compatible or custom payloads |
Webhook sources do not currently use Shankh's native discovery, automatic alert setup or drift checking.
Connect a Vörr connector
- Open Vörr.
- Select Add connectors.
- Connect the required source and complete its configuration.
- Return to Shankh and confirm the source is available in your alerting setup.
Enter AWS alert-management credentials
- Open Cloud Integration.
- Enter the alert-management credentials for the AWS account.
- Save the credentials.
- Confirm the account's resources become available for selection.
Permissions vary by provider. Follow the permissions explicitly documented for your provider rather than a generic minimum set.
The credentials are stored encrypted in Frigga's database.
Choose the cloud, service and metrics
- Open the cloud account in Cloud Integration.
- Check the services Shankh discovered on it.
- Open a service to get its available metrics.
- Enable the metrics that should produce alerts.
- Set the threshold for each severity.
- Review the workspace assignment.
- Preview the alert configuration.
- Deploy the configuration.
Only the metrics the cloud actually exposes for that service are offered — the list comes from the account, not from a fixed catalogue.
Set a severity threshold
- A threshold can include metric, operator, value, severity, evaluation period, datapoints required to alarm, and statistic or percentile where supported.
- Use values that match the service and environment, not values copied from an unrelated service.
- Make the evaluation window long enough to survive normal spikes.
- Match the severity to how urgently a human needs to act.
- Select the metric to evaluate.
- Set whether a higher or lower value is a breach.
- Enter the threshold value.
- Assign the severity for that breach.
- Set the evaluation window.
- Set the number of periods that must breach before an alert fires.
- Save and deploy.
Verify the deployment
- Confirm the deployed configuration appears against the service.
- Confirm the workspace assignment matches the team that owns the service.
- Trigger or wait for a real breach, then confirm an incident appears in Incidents.
If no incident appears, work through the end-to-end test.