You can create alerts to monitor CPU usage in your Kubernetes setup and get notified when containers exceed specified thresholds. This guide shows you how to set up an alert that triggers when CPU usage goes above 80%.
Prerequisites
- SigNoz installed and configured
- Kubernetes cluster with kubeletstats receiver configured
- The
k8s.container.cpu_limit_utilizationmetric enabled (enabled by default in k8s-infra Helm chart v0.11.10+)
Setting up the CPU usage alert
Follow these steps to create an alert for container CPU usage exceeding 80% of its limit:
Step 1: Create the metric query
Navigate to Alerts in the SigNoz UI and create a new alert with the following configuration:
- Metric:
k8s.container.cpu_limit_utilization - Time aggregation:
avg - Space aggregation:
maxby (k8s.cluster.name,k8s.namespace.name,k8s.pod.name,k8s.container.name)
This configuration:
- Uses
avgtime aggregation to smooth out short CPU spikes over the evaluation window - Uses
maxspace aggregation to alert when any container in a pod exceeds the threshold - Groups by cluster, namespace, pod, and container names for granular monitoring
Step 2: Set alert condition
Configure the alert evaluation:
- Evaluate: Query A
- Condition: Above
- Threshold:
0.8(represents 80%) - Match type: all the times
The threshold of 0.8 corresponds to 80% because k8s.container.cpu_limit_utilization is a ratio metric (0.0 to 1.0).
Expected result
This configuration will monitor all containers across your Kubernetes clusters and trigger an alert whenever any container's CPU usage exceeds 80% of its configured limit. The alert will include the cluster name, namespace, pod, and container name in the notification, making it easy to identify which container needs attention.
Next steps
- Learn more about Kubernetes metrics
- Explore time aggregation best practices for alerts
- Set up notification channels to receive alerts via Slack, PagerDuty, or other channels
- Create similar alerts for memory usage using
k8s.container.memory_limit_utilization
