Latest from Abhimanyu's Blog | Scaling infrastructure, one cluster at a time

Pods can now exclude tainted nodes during topology spread calculations, improving placement predictability.

Kubernetes 1.33 ensures PV reclaim policies are honored even if PVs are deleted before PVCs, preventing storage leaks across CSI and in-tree drivers.

A Kubernetes pod takes a fast async path, skipping blocking API calls handled in the background, showing the shift from sync to async preemption.

Kubernetes adds limited swap support for Burstable pods, offering memory flexibility on cgroupsv2 nodes without compromising workload stability.

Kubernetes now aligns memory-backed emptyDir volumes with pod memory limits for improved portability and consistency across node types.

Kubernetes wasn’t built from scratch. Learn how Google’s secret systems shaped its design, and why that origin still matters for developers today.

Use Pluto to identify deprecated or removed Kubernetes APIs in your manifests and Helm charts before upgrading, ensuring smooth and predictable cluster upgrade.

A new /statusz endpoint is coming to Kubernetes. Find out how it boosts debugging and observability without touching your metrics stack.

A critical kubelet bug exposes a DoS risk via the unauthenticated /checkpoint API. Learn how to detect, mitigate, and patch CVE-2025-0426.

Kubernetes v1.33 lets you configure container stop signals via PodSpec, no more rebuilding images just to change shutdown behavior.

Discover how Kubernetes v1.33 introduces a new /flagz endpoint in Kubelet for runtime introspection of component flags, debug like never before.

Kubernetes 1.33 speeds up recovery with a 1s initial delay and 60s max backoff for restarts, opt-in via feature gate for faster handling of failing containers.

Pods that grow with your workload? Discover how Kubernetes v1.33 lets you scale CPU and memory without a restart, and when it still might not be enough.

Kubernetes v1.33 finally enforces image pull secrets even for cached images, closing a 10-year-old loophole in multi-tenant cluster security.

Did you know you can recover deleted Kubernetes resources from etcd snapshots without downtime or cluster rollback? Most don’t, it’s surprisingly simple.

Helm upgrade failed after moving to Kubernetes v1.25? Here’s how I fixed it by cleaning up legacy PDB API references using helm-mapkubeapis.

Schedule pod downscaling in Kubernetes with kube-downscaler to cut costs during off-hours—my experience, setup, and where it fits best.

A 2025-ready Kubernetes production checklist you can’t afford to miss, real-world tools, CNCF-backed standards, and proven strategies now featured by CNCF.

Overwhelmed by Kubernetes? Don’t get left behind. Use these tips to simplify scaling, improve security, enhance monitoring, and streamline daily operations.

Kubernetes v1.33 adds per-HPA configurable tolerance, allowing fine-tuned scaling sensitivity for both scale-up and scale-down decisions.

JobSet is a Kubernetes-native API for managing distributed ML and HPC jobs with support for multi-role pods, topology-aware placement, and scaling.