Epsagon alerts support many use cases: Kubernetes metrics, application performance metrics, and application exceptions.

🚧

Exception alerts

Exception alerts are collected from the tracing libraries and available only for traced services and code

Manage your alert rules

• To manage your alert rules, go to Alerts
• To create a new alert rule, click Create New Alert.
  To set up an alert for exceptions, go to the alerts page, and create a new "simple" alert. In the alert type - choose Exception.

Lambda Events

Lambda event alerts are triggered for each Lambda invocation event of specific types.

• Select event types to trigger the alert. Support events types are: Timeout, Out of Memory, Code Exception (available for traced functions), Function Error (identified by CloudWatch logs), Insight (function is close to the time limit or memory limit).
• Filter which Lambda functions will trigger the alert by Application, specific Functions or AWS Accounts.

Lambda Metric

Lambda metric alerts are triggered whenever Lambda invocations cross a specific threshold over time. For example, when the average number of invocations ending in timeout is greater than 10 invocations for a period of 15 minutes.

To create a Lambda metric alert:

• Select an aggregation function
• Select the invocation event on which to base the metric
• Select an operator and the threshold value
• Select the timeframe in which to check the metric value before triggering the alert

Kubernetes Alerts

Kubernetes alerts trigger when a Kubernetes metric crosses a threshold. For example, when the average CPU usage of all pods is above 95% for 5 minutes.

To create a Kubernetes alert:

• Select which Kubernetes entity you want to monitor: Node, Pod, or Container
• Select a metric to check
• Select an operator and threshold value
• Select a timeframe in which to check the metric value before triggering the alert
• Select which entities to check based on Deployment, Namespace or Cluster.

Trace Metrics Alerts

Trace metric alerts trigger when a check on trace data is met. For example, when the average duration of putting items in a specific DynamoDB is longer than 2 seconds.

To create a trace metric alert:

• Select an aggregation function
• Select the metric you want to check. Possible metrics are indexed numeric trace tags.
• Select an operator and threshold value
• Filter traces to match the use case you want to check. For example, filter traces by a Kafka stream name, or Kubernetes cluster.