Kube Builders

KSail is a unified Kubernetes SDK that simplifies cluster management and workload deployment by providing a single, declarative interface for all your Kubernetes operations. More: https://ku.bz/bVlZGXZd7

This tutorial shows how to connect on-prem Kubernetes workloads to Google Cloud without service account keys using Workload Identity Federation, OIDC, Terraform, Kyverno, and IAM attribute conditions. More: https://ku.bz/1YVD6c3FP

KEDA GPU Scaler is a KEDA external scaler that reads NVIDIA GPU metrics through NVML and autoscalers vLLM, Triton, training jobs, and custom inference workloads without requiring Prometheus. More: https://ku.bz/Hg61Tjm7h

In this article, you will learn how to run MySQL on Kubernetes in production, including deployment topologies, storage choices, monitoring, backups, failover, and operator selection. https://ku.bz/DmXPD42-9

This tutorial teaches how to build a local observability stack for AWS EKS logs using Stern for multi-pod tailing, Fluent Bit for log processing with multiline parsing, and Elasticsearch with Kibana for searchable visualization via Docker Compose. More: https://ku.bz/qK6fTw_SB

"Humans like to look at those things so that they can build confidence." Yasmin Rajabi explains how CloudBolt Software is turning survey learnings into product guardrails: more visibility into AI/ML recommendations, controls over when automation runs, and clearer failure-handling paths. The product challenge is building trust before automation changes production resources. Watch the announcement: https://ku.bz/HDtVsM95b Read The Kubernetes Automation Trust Gap study: https://ku.bz/449hgHFbV

Dave Masselink, Software Engineer and Founder at Compute Gardener, explains how time-of-use electricity pricing works and how it can be leveraged for cost optimization in Kubernetes workloads. He breaks down utility rate structures where customers pay different rates during peak (20 cents/kWh) versus off-peak periods (10 cents/kWh), compared to standard flat rates (15 cents/kWh). The discussion highlights the practical scheduling strategy of working with fixed utility schedules and the growing relevance of cost-aware computing as cloud providers begin advertising their carbon intensity profiles. Watch the full episode: https://ku.bz/zk2xM1lfW

EgressGateway provides stable egress IP addresses for pods accessing external services. It works across CNI networks like Calico and Flannel, offering high availability, fine-grained policy control, and IPv4/IPv6 dual-stack. More: https://ku.bz/1zqlTQGdD

Upgrading your OS shouldn't break your cluster — but when your GPU stack is tightly coupled to the operating system, it gets complicated. Jordan Karapanagiotis runs K3s on NVIDIA Jetson devices using JetPack, which is based on Ubuntu. Upgrading NVIDIA's toolkit means upgrading the entire OS. The cluster itself stays the same, but the networking layer is where things get fragile — connect to a hotspot instead of a stable network, and things can completely break. When your infrastructure isn't a typical cloud server, you have to build your Kubernetes cluster to be robust against changes to the environment underneath it. Watch the full interview: https://ku.bz/3yDmJ_xSC

A storage-only dashboard is a dashboard nobody looks at. Amin Astaneh from Certo Modo makes the case for service level objectives over raw metrics: start with how your application is performing, then use the USE Method (Utilization, Saturation, Errors) to explain why. Storage metrics are a diagnostic tool, not a destination. Don't build dashboards for the sake of it — build dashboards that answer questions. Watch the full interview: https://ku.bz/4CbNjNMRL

git-change-operator is a Kubernetes operator that enables automated Git operations from within clusters through GitCommit and PullRequest custom resources. More: https://ku.bz/Y1q8PFFvw

Orphan Resource Collector (ORC) is an open-source tool for detecting and managing orphaned resources in Kubernetes clusters. More: https://ku.bz/yd4ZdWCN8

This week on Learn Kubernetes Weekly 190: 🌪️ Taming the Storm: Building Groww's Internal Chaos Engineering Platform 🧪 Taracode Testing a Go-Based CLI AI Agent in My Homelab 🧠 Building self-evolving AI systems: exploring the architecture 🔄 Migrating from slurm to Kubernetes 🏠 Lessons Learnt Self-hosting an AI Assistant Read it now: https://kube.today/issues/190 ⭐️ This newsletter is brought to you by LearnKube — master Kubernetes with hands-on training designed for engineers who want to learn the smart way https://ku.bz/hypSbyc-V

This article teaches how to build self-evolving AI systems using Kubernetes, Temporal workflows, and automated deployment pipelines, enabling AI agents to detect errors, fix code, and redeploy services without manual intervention. More: https://ku.bz/CqL64wJNt

This tutorial shows how to serve open source LLMs on Red Hat OpenShift AI with KServe, vLLM, Argo CD, and IBM Fusion using a fully declarative GitOps workflow. More: https://ku.bz/chQc0HG6n

New on LearnKube: Server-side apply: what happens when you run kubectl apply Server-side apply changes how Kubernetes handles field ownership. Kubernetes objects are shared state. Manifests, controllers, release tools, autoscalers, webhooks, and operators can all shape the same object. With client-side apply, stale intent can overwrite live changes. With server-side apply, ownership moves into the API server. Kubernetes tracks which manager owns each field and surfaces conflicts when ownership is contested. Chiara put serious work into this guide, and you can read it in full here: https://learnkube.com/server-side-apply-kubernetes

GPU optimization requires application-level thinking, not just hardware metrics. Amin reminds us that HPC fundamentals apply: golden signals like throughput and latency are critical for expensive hardware. Don't fixate on utilization alone - application performance metrics matter more. Watch the full interview: https://ku.bz/p1RNM5ldZ This interview is a reaction to Dave Masselink's episode https://ku.bz/zk2xM1lfW

Goldpinger is a monitoring tool that runs as a DaemonSet and makes calls between pod instances to test connectivity. It produces Prometheus metrics for visualization and alerting, while providing a web UI that displays network health and latency. More: https://ku.bz/WD6zf4B2h

This tutorial shows how to build a private EKS cluster with zero public API exposure using Terraform. It also covers self-hosted OpenVPN as a VPN gateway, NAT masquerade iptables setup, kube-prometheus-stack via internal load balancer, and Route 53. More: https://ku.bz/XZ-gvhGdK

Kube Binpacking Exporter exposes Prometheus metrics that show how efficiently your cluster packs requested CPU and memory across nodes, groups, and DaemonSet overhead so you can measure fragmentation over time. More: https://ku.bz/VpnFlkz-0