https://assets.learnk8s.io/linkedin-179.png
This week on the Learn Kubernetes Weekly:
☁️ Event-Driven Architecture Missing Standards
🔥 Sandboxes for AI
🔐 Securing East-West Traffic
💥 Chaos Engineering Best Practices
📊 AWS Monitoring Platform
⭐️ Portworx
Read it now: https://kube.today/issues/179
This article shows that default Kubernetes system pods such as kube-proxy and CoreDNS ship with incomplete or missing memory configurations, making them vulnerable to OOM kills during memory pressure
➜ https://ku.bz/14Tn3Xwvk
https://kube.careers/image-gen/digest?companies=Pinterest&companies=NVIDIA&companies=Perplexity¤cy=USD&salaryFrom=180000&salaryTo=499541&subtitle=This+week&title=6+Kubernetes+Platform+Engineer+jobs
https://miro.medium.com/v2/1*CvDU1W672ulF0Vv8odJVuA.png
This article explains service mesh patterns for managing microservice communication, covering how sidecars like Envoy handle retries, circuit breakers, timeouts, and load balancing transparently
➤ https://ku.bz/CZy4X93Wv
MCP Operator deploys and validates MCP servers on Kubernetes with automatic protocol detection for SSE and Streamable HTTP, built-in monitoring via Prometheus and Grafana dashboards, optional metrics sidecar for request tracking, and HPA support
➜ https://ku.bz/7GhL57BxZ
https://kube.events/image-gen/digest?ids=d54782ba-1a15-4027-9cfd-a95607671024,e6bd1714-9ff2-476a-82f0-a2d6abb3f88d,99ee98f0-6842-46ab-94ca-14f03617e6a2,734d737b-7bbb-4221-97e8-1c200230a81a,2cb6c6e1-dd6b-4ded-b882-ddf2132dce26,d4fd336f-31b1-4ff1-ad57-3eebe4f70638,7db02014-f134-4b17-84f2-41bc4c028758
https://res.cloudinary.com/learnk8s/image/upload/v1776067746/controller-manager-explained-2026/thread-slide-1.png
Delete a Pod from a Deployment and a new one appears.
The API server didn't create it. The scheduler didn't schedule it yet. A controller inside the controller manager noticed the count was wrong and created a replacement before anything else happened.
https://github.com/containerd/containerd/raw/81bc6ce6e9f8f74af1bbbf25126db3b461cb0520/docs/cri/architecture.png
zeropod is a tool that automatically checkpoints containers to disk after a certain amount of time of the last TCP connection, allowing for fast and seamless scaling down to zero
➤ https://ku.bz/MjQ_WWsrx
K8sQuest is a local Kubernetes learning game with 50 progressive challenges where you fix broken clusters using kubectl with real-time monitoring, progressive hints, and post-mission debriefs running on kind
➤ https://ku.bz/j4c05Xsk0
https://github.com/dgunzy/flux9s/raw/main/docs/images/filter-screenshot.png
flux9s is a terminal interface for monitoring Flux GitOps resources in real time with K9s inspired navigation
It offers resource operations like:
- suspend,
- resume,
- reconcile,
- graph visualization,
- reconciliation history tracking
➤ https://ku.bz/jvKh9NH_1
https://miro.medium.com/v2/1*sSYdrTz9FloYet8ZYI8A-Q.png
This article explains why Linecraft AI migrated from manual Windows IIS deployments to Kubernetes after facing multi-week release cycles, inconsistent environments from local server changes, and extended downtimes
➤ https://ku.bz/vGgfxBD25
https://miro.medium.com/v2/1*wgGSKOqeaYBkSXYy35g41g.png
This tutorial teaches how to implement multi-tenancy on Google Kubernetes Engine using namespaces for isolation, RBAC for access control, and resource quotas for capacity management
➤ https://ku.bz/Qd8k933tW
https://kube.careers/image-gen/digest?companies=Galileo+Financial+Technologies&companies=Palantir+Technologies&companies=Freedom+Technology+Solutions+Group¤cy=USD&salaryFrom=105600&salaryTo=198000&subtitle=This+week&title=6+Kubernetes+junior+jobs
HAProxy Data Plane API is a sidecar process written in Go that runs alongside HAProxy and provides REST API endpoints for managing HAProxy configuration, requiring HAProxy version 1.9.0 or higher
➤ https://ku.bz/wyGW4WtdG
https://miro.medium.com/v2/1*itKLepsppxnGUw68iC7YtQ.png
This article explains chaos engineering practices for Kubernetes, covering how to design resilient systems by proactively testing failure scenarios like pod crashes, network failures, and resource exhaustion using tools like Chaos Mesh and LitmusChaos
➜ https://ku.bz/y74YwPMbq
"Before Hadron, we were retrofitting general-purpose OS. Now we have a minimal system designed exclusively for edge workloads."
Ettore Di Giacinto on the shift
📺: https://ku.bz/wMhKpZ5bQ
📣: https://ku.bz/_9RmXnjDJ
Linnix is an eBPF + PSI-powered Kubernetes observability agent written in Rust that identifies which pod is actually stalling your services, not just consuming CPU
➤ https://ku.bz/x-VQLHwSW
This article shares hidden Kubernetes bad practices discovered through production incidents covering troubleshooting challenges, configuration mistakes, and operational lessons learned by a Site Reliability Engineer
➤ https://ku.bz/5r3LtnjD5
This tool exposes Kubernetes cluster operations through the Model Context Protocol, allowing AI agents and tools to safely read cluster state and execute controlled actions like kubectl commands
➜ https://ku.bz/cY99-lPN-
https://miro.medium.com/v2/0*XIeJGF7iQrF91hUn.png
This case study describes rebuilding a fragile Kubernetes infrastructure into a production-grade platform for GPU-based ML workloads, improving deployment frequency from weekly to 10+ times daily
➤ https://ku.bz/lFNBx-8yT
Routing. Engine. Caching. Orchestration. Nodes. That's RECON — a framework for the AI inference stack
Tsahi Duek on the five layers every AI platform needs
📺: https://ku.bz/2r41YKBZb
This article introduces Gt and Lt operators for pod tolerations in Kubernetes 1.35, enabling threshold-based scheduling with numeric taint values, such as failure probability or cost metrics, rather than exact matching
➜ https://ku.bz/vW6JV-99T
🗣️ Tanat Lokejaroenlarb explains why EKS in-place upgrades are far more complex than the "one-click" perception many have
Watch the full episode: https://ku.bz/VVHFfXGl_
https://assets.learnk8s.io/linkedin-178.png
This week on the Learn Kubernetes Weekly:
🔥 RCE via Nodes/Proxy
🦅 Chaos to Engineering Excellence
☸️ Extended Toleration Operators
🔄 K8S to ECS Fargate
🗄️ SQL Server on AKS with GitOps
⭐️ StormForge
Read it now: https://kube.today/issues/178
Stakpak is an open source terminal AI agent for DevOps that generates infrastructure code, debugs Kubernetes, and automates deployments — with secret substitution so the LLM never sees your actual credentials
➜ https://ku.bz/xXr0Rbc2c
KubeProxy routes once per connection, not per request. For gRPC, some pods burn at 80% CPU while others sit idle
Rohit Agrawal from Databricks on replacing it with proxy-less, client-side load balancing
https://ku.bz/y803JMhBk
https://kube.careers/image-gen/digest?companies=Pinterest&companies=NVIDIA&companies=Perplexity¤cy=USD&salaryFrom=224000&salaryTo=499541&subtitle=This+week&title=6+Kubernetes+Platform+Engineer+jobs
https://github.com/sgl-project/rbg/raw/main/doc/rbgs-concept.png
RoleBasedGroup is a Kubernetes API written in Go for orchestrating distributed stateful AI inference workloads with multi role collaboration and built in service discovery, treating inference services as role based groups rather than isolated workloads
➤ https://ku.bz/-sFQ3PwFR
This case study shows how NetworkLessons migrated from Kubernetes to ECS Fargate using AWS CDK, reducing operational complexity while implementing multi-account architecture, automated cost controls, and infrastructure as code
➜ https://ku.bz/T_hNMhdc4
https://miro.medium.com/v2/0*y_Nt0_k_wPBM3aQ5.png
This tutorial teaches how to set up a local DNS server specifically for demo environments using dnsmasq and Docker containers
➜ https://ku.bz/r6rbLZ-dH
