Kubernetes v1.36 Introduces PSI Metrics to Enhance Resource Monitoring

Published on May 13, 2026

Monitoring resource usage in Kubernetes clusters has typically relied on standard metrics like CPU and memory utilization. However, these metrics alone can be misleading and insufficient for effective resource management. The introduction of Pressure Stall Information (PSI) metrics in Kubernetes v1.36 marks a significant shift in how users can assess resource contention across their nodes and containers.

The recent rollout of Kubernetes v1.36 brought PSI metrics out of beta and into General Availability (GA). This change allows users to access more precise data regarding stalled processes and their causes. As Kubernetes environments become increasingly complex, the new metrics provide vital insights into potential bottlenecks before they escalate into critical outages.

Evaluation of performance showed that integrating PSI does not impose significant overhead on cluster resources. Tests validated that with high-density workloads, the Kubelet’s ability to collect and surface PSI metrics operates efficiently within existing resource usage patterns. This makes it safe and practical for production environments, offering operators a reliable way to monitor real-time resource conditions.

The impact is already being felt across the Kubernetes ecosystem. Users can now proactively respond to resource contention, potentially avoiding outages and maintaining overall system health. As PSI metrics offer a clearer picture of resource dynamics, teams can optimize their deployments, enhancing resilience and performance.

Related News