Unlocking Kubernetes v1.36: PSI Metrics for Proactive Resource Management
Kubernetes v1.36 brings PSI metrics into general availability, addressing a critical need for proactive resource management. PSI provides high-fidelity signals that help you identify resource saturation before it leads to an outage. This capability is essential for maintaining application performance and reliability in production environments.
The Kubelet now detects OS-level PSI support through cgroup configurations, ensuring that pressure metrics are only collected when supported by the node. This means cleaner data for your monitoring and alerting systems. To take advantage of these metrics, ensure your nodes are running a Linux kernel version 4.20 or later and using cgroup v2. Additionally, your kernel must be compiled with CONFIG_PSI=y, and the system must not be booted with the psi=0 parameter.
In production, you can query PSI metrics using a simple command. For example, use kubectl get --raw to access the stats summary for a specific container. Be cautious, as proxying to the kubelet is a privileged operation that requires appropriate administrative permissions. As of v1.36, you no longer need to opt in to any feature gate, making it easier to leverage these metrics in your workflows.
Key takeaways
- →Leverage PSI metrics to identify resource saturation before outages occur.
- →Ensure your kernel is compiled with CONFIG_PSI=y to collect accurate PSI data.
- →Use moving averages (10s, 60s, 300s) to differentiate between transient spikes and sustained resource tension.
- →Query PSI metrics with `kubectl get --raw` for real-time insights into container performance.
- →Be aware of security risks when proxying to the kubelet; use appropriate permissions.
Why it matters
In production, PSI metrics can significantly reduce downtime by allowing you to anticipate and address resource issues before they impact users. This proactive approach enhances overall system reliability.
Code examples
CONTAINER_NAME="example-container"
kubectl get --raw "/api/v1/nodes/$(kubectl get nodes -o jsonpath='{.items[0].metadata.name}')/proxy/stats/summary" | jq '.pods[].containers[] | select(.name=="$CONTAINER_NAME") | {name, cpu: .cpu.psi, memory: .memory.psi, io: .io.psi}'When NOT to use this
The official docs don't call out specific anti-patterns here. Use your judgment based on your scale and requirements.
Want the complete reference?
Read official docsIndustry-standard certifications built by the people behind Linux and Kubernetes. Earn the CKA — the gold standard Kubernetes administrator cert. OpsCanary readers get 30% off year-round with code OPSCANARY3.
Get CKA certified →Dynamic Resource Allocation in Kubernetes: Does It Replace HAMi?
Dynamic Resource Allocation (DRA) is now a core feature in Kubernetes, but does it truly replace HAMi? With DRA's claims model and consumable capacity, understanding the nuances is crucial for effective resource management.
Harnessing CoHDI: Transforming Kubernetes with Composable Disaggregated Infrastructures
Kubernetes is evolving, and CoHDI is at the forefront of this transformation. With Dynamic Resource Allocation (DRA), it allows for the dynamic attachment and detachment of PCIe devices, paving the way for more efficient resource management in your clusters.
Debugging GPU Utilization in Kubernetes: The Cilium and Kubeflow Challenge
Struggling with idle GPUs in your Kubernetes cluster? Discover how Cilium's topology-aware networking can clash with Kubernetes scheduling, leading to inefficient resource use. Learn how to leverage nodeAffinity and podAffinity to optimize your GPU workloads.
Get the daily digest
One email. 5 articles. Every morning.
No spam. Unsubscribe anytime.