DevOps & SRE notes

The article explores how integrating these two prominent tools can accelerate and enhance platform engineering initiatives. https://platformengineering.org/blog/platform-engineering-with-crossplane-and-argocd

Open, Multi-Cloud, Multi-Cluster Kubernetes Orchestration https://github.com/karmada-io/karmada

A userspace out-of-memory killer https://github.com/facebookincubator/oomd

без подписи

Hands-on comparison against a deliberately broken cluster, with real outputs and failure-mode differences. Practical for deciding where AI Kubernetes tools fit: scanner, agent framework, or natural-language kubectl layer. https://decodeops.substack.com/p/k8sgpt-vs-kagent-vs-kubectl-ai-what

Deep-dive technical guide into the exact mechanics of pod restarts and configuration updates in Kubernetes https://www.cncf.io/blog/2026/03/17/when-kubernetes-restarts-your-pod-and-when-it-doesnt/

kubectl debugging plugin to collect full or partial cluster state and serve via an api server. Kubernetes time machine https://github.com/crust-gather/crust-gather

The 10-step playbook to reduce K8s bills by 40-60%, focusing heavily on bridging the gap between requested and actually used resources. https://leanopstech.com/blog/kubernetes-cost-optimization-guide-2026/

Ministack: Free, open-source local AWS emulator - 55+ services, Terraform compatible, real databases. Free forever. MIT licensed. https://github.com/ministackorg/ministack

Containerlab focuses on the containerized Network Operating Systems which are typically used to test network features and designs https://github.com/srl-labs/containerlab

Deep dive into Prometheus’s use-uncached-io work and why page cache behavior can make Kubernetes container memory metrics misleading. Useful for anyone running Prometheus at scale: covers memory predictability, compaction writes, OOM risk, and tradeoffs around direct I/O. https://prometheus.io/blog/2026/03/05/uncached-io/

Kubernetes configuration tracking controller. Wave watches Deployments, StatefulSets and DaemonSets within a Kubernetes cluster and ensures that their Pods always have up to date configuration. By monitoring mounted ConfigMaps and Secrets, Wave can trigger a Rolling Update of the Deployment when the mounted configuration is changed. https://github.com/wave-k8s/wave

без подписи

Lessons from Moving a Live Production Database describes the complex process of migrating a massive dataset while keeping the service available to users. The case study covers specific strategies for achieving zero downtime during a high-risk infrastructure change. But is zero downtime always worth the engineering effort? Imagine that you have two options: 1. Spend several weeks preparing and testing a zero-downtime migration. 2. Schedule 20 minutes of downtime during a low-traffic period. Considering that 99.9% availability allows approximately 43 minutes of downtime per month, which option would you choose? What factors would change your decision: revenue loss, SLA penalties, customer expectations, rollback complexity, or the size of the engineering team? https://www.tines.com/blog/zero-downtime-database-migrations-lessons-from-moving-a-live-production/

A lightweight tool for deploying and managing containerised applications across a network of Docker hosts. Bridging the gap between Docker and Kubernetes ✨ https://github.com/psviderski/uncloud

без подписи

Hands-on vendor-neutral instrumentation for GenAI spans, tool calls, token metrics, and trace exploration. Worth reading if you want AI-agent debugging to fit existing OTel/Grafana/Loki-style observability rather than a separate black box. https://opentelemetry.io/blog/2026/genai-observability/

Trusted builds made easy! A cloud-native software factory for building, testing, and releasing trusted software artifacts Konflux is an open-source, Kubernetes-native CI/CD platform that manages the full software delivery lifecycle for software artifacts — with supply chain trust built in from the start. Built on Tekton and the Conforma policy framework, it brings together best-in-class open source projects into a single, integrated software factory. Managed by a Kubernetes operator, Konflux runs on Kind, OpenShift, and any conformant Kubernetes cluster. https://github.com/konflux-ci/konflux-ci

Morgan Stanley presented their journey scaling Flux to manage over 500 clusters and tens of thousands of resources. They detailed their transition from push-based CI/CD pipelines to a pull-based, continuous reconciliation model using Flux to eliminate configuration drift. The article outlines their strategy for handling enterprise-grade security, performance tuning for scale, and adapting architecture (like moving from Git to S3) to meet high availability and compliance needs. https://fluxcd.io/blog/2026/03/stairway-to-gitops-morgan-stanley/

https://www.cncf.io/blog/2026/06/25/building-a-cluster-aware-ai-agent-with-kubernetes-argo-cd-and-gitops/ Practical walkthrough for a self-hosted, read-only cluster AI agent with GitHub Actions, Argo CD Image Updater, and no cluster data leaving the network. Good pattern material for safe “AI assistant inside the cluster” experiments.