Configure New Relic NRQL alert conditions for error rate and latency SLO thresholds

Steps

1. Create an alert policy via the NerdGraph alertsPolicyCreate mutation or the New Relic UI; specify the incident preference (PER_POLICY, PER_CONDITION, or PER_CONDITION_AND_TARGET) to control how multiple triggered conditions combine into incidents.
2. Add a NRQL alert condition using the alertsNrqlConditionStaticCreate mutation; write a NRQL query returning a single numeric value per evaluation (e.g., SELECT percentage(count(*), WHERE error IS true) FROM Transaction WHERE appName = 'MyApp'); set the aggregationMethod to EVENT_FLOW or EVENT_TIMER depending on streaming data characteristics.
3. Set the threshold as a critical value (e.g., error rate > 1%) with a threshold duration expressed in minutes (e.g., for at least 5 minutes); optionally add a warning threshold at a lower severity level.
4. Configure the expiration policy to specify how the condition behaves when data stops arriving (closeViolationsOnExpiration, openViolationOnExpiration); for SLO alerting, avoid opening a violation on data expiration to prevent false alerts during deployments.
5. Attach a notification channel (Slack, PagerDuty, email) to the alert policy via the alertsNotificationChannelCreate mutation or Workflows (the preferred modern approach) using alertsWorkflowCreate to route alerts with enrichment from NRQL queries.
6. Test the alert condition with the preview feature in the New Relic UI (signal history chart) before activating; verify that historical data would have triggered alerts at the correct times relative to known incidents.