Description
When GKE Cloud Monitoring integration is disabled, required cluster-health and resource-usage metrics may be unavailable. Components and metrics are selected separately; enabling integration does not automatically configure every application metric or alarm.
Potential impact
- Resource shortages or performance degradation may be noticed late.
- Troubleshooting and capacity planning may lack necessary evidence.
Remediation
- Enable required metrics with the supported
monitoring_service = "monitoring.googleapis.com/kubernetes"setting or current monitoring configuration. Use supported settings instead of the retired legacy integration. - Verify actual metric receipt and collection permissions, and configure required alarms. Review collection scope and cost.
Examples
These excerpts show part of a Standard cluster configuration. Supply the omitted network and node settings for your environment.
Before
hcl
resource "google_container_cluster" "cluster" {
name = "marcellus-wallace"
location = "us-central1-a"
initial_node_count = 3
monitoring_service = "none"
}
After
hcl
resource "google_container_cluster" "cluster" {
name = "marcellus-wallace"
location = "us-central1-a"
initial_node_count = 3
monitoring_service = "monitoring.googleapis.com/kubernetes"
}
Explanation:
- Before: Monitoring integration is disabled. Check whether another monitoring system meets requirements.
- After: The current GKE Monitoring integration is selected. Verify required metrics and alarms separately.