Custom evaluation delay
Overview
Lags in your data pipeline can cause delays in log and metric ingestion, potentially leading to false alerts. While alert conditions are evaluated in real-time, delayed data arriving later can retroactively affect whether those conditions are met.
When an alert should be triggered above a user-defined threshold, missing data may delay it or lead to an incorrect resolution. Conversely, when an alert should be triggered below a user-defined threshold, incomplete data can cause a false positive alert or delayed alert resolution.
The Custom Evaluation Delay feature helps mitigate this issue by shifting an alert's evaluation timeframe backward by a configurable amount. This adjustment ensures that alert conditions are evaluated against a more complete dataset, accounting for late-arriving logs or metrics.
By using Evaluation Delay, you can reduce the risk of false positives or negatives caused by real-time data fluctuations and improve the accuracy and reliability of your alerts.
Configuration
- Open the alert creation wizard: go to Alerts, then select Create alert. You can also start from a saved query in Explore and select Create alert.
- In the Condition step, expand Advanced configurations.
- Set Delay alert evaluation to the time period you want the evaluator to wait for late-arriving data.
Related resources
Next steps
Adjust anomaly detection thresholds for your alerts in Customizing anomaly detection alert sensitivity.