Kubernetes v1.34 has promoted Pressure Stall Information (PSI) metrics to Beta, offering a critical tool for monitoring node health in growing clusters. PSI, a Linux kernel feature, quantifies resource pressure by measuring task stall time due to contention, rather than just utilization. It provides metrics for CPU, memory, and I/O, categorized as "some" pressure (at least one task stalled) or "full" pressure (all non-idle tasks stalled). These metrics are available over rolling windows of 10 seconds, 1 minute, and 5 minutes. The KubeletPSI feature gate allows the kubelet to collect and expose PSI metrics through the Summary API and the /metrics/cadvisor Prometheus endpoint. New Prometheus metrics like container_pressure_cpu_stalled_seconds_total are now accessible. These metrics help identify memory leaks, optimize resource requests, and trigger autoscaling. To enable PSI, nodes must run Linux kernel 4.20+ with cgroup v2, and the KubeletPSI feature gate must be enabled. PSI metrics are Linux-specific and not available on Windows nodes. This beta feature is being actively developed, and feedback is encouraged.