Learn Kubernetes Weekly issue 205 · 14 Oct 2026
Kubernetes Health Checks Down Payments, Bare-Metal Hybrid Surveys, No Redis Calls, CNI Deep Dive, Why CPUs Never Had This Problem
This newsletter is brought to you by LearnKube — understand how Kubernetes works, and what to do when it breaks. Live training with 60% hands-on labs.
Do you know about the LearnKube Telegram channel?
Recently, we shared a case study about a payment service stuck in a restart loop. Its liveness probe depended on a downstream service.
Someone asked: can restarts really push containers into a resource-constrained state?
One possible explanation is CPU competition at startup. Imagine ten containers that each need a full CPU to start, but share a node with four CPUs. If they restart together, startup takes longer. Probe responses can take longer too, which can trigger another restart.
Then came the next question: if the container needs all that CPU at startup, how did it work in the first place?
That second question is a good one. It makes you consider what changed on the node, even when the application configuration stayed the same.
A container can use spare CPU beyond its request if its CPU limit permits it. But spare CPU is a different story when all ten containers need it at once.
If you read that and have another question, I want to hear it.
And if you want to explore Kubernetes with us over four days, our next online Advanced Kubernetes course starts next Thursday, October 22.
The dates and full agenda are here →
Now, on to this week’s articles!
— Dan
Articles
How Kubernetes Health Checks Brought Down a Payment Service
medium.com
This case study shows why a Kubernetes liveness probe must not depend on downstream services after one caused payment pods to restart repeatedly, and how separate readiness and liveness checks avoid the same failure.
The cluster that runs on our own metal: rebuilding survey processing on hybrid Kubernetes
medium.com
This case study shows how Propeller replaced its custom scheduler with a hybrid Kubernetes cluster that runs GPU jobs on its own servers and uses AWS for extra capacity.
The Fastest Redis Call Is the One You Never Make
engineering.gokwik.co
This case study shows how GoKwik added an in-process cache to every pod, used Redis keyspace events and TTLs to clear stale data, and cut Redis network traffic by about 72%.
newtonwamiti.substack.com
This article explains how CNI connects Kubernetes pods, how ADD, DEL, and CHECK work, and how Flannel, Calico, and Cilium move traffic differently.
Why did CPUs never have this problem?
medium.com
This article explains why common GPU utilization numbers can hide idle compute and how allocation, active time, and useful model work reveal different kinds of GPU waste.
Container-Aware Resource Management in Go: The Problem, Go 1.25’s Fix, and What’s Still Missing
dev.to
This article explains how Go 1.25 makes GOMAXPROCS respect container CPU limits, why GOMEMLIMIT still needs manual setup, and how you can prevent CPU throttling and OOM kills in Kubernetes.
Articles worth checking out:
4 Days Advanced Kubernetes Course
Deploy your app to Kubernetes without fear.
In 4 days, you'll cover Helm & Kustomize, zero-downtime deployments, networking & Ingress, autoscaling, and security, with 60% hands-on labs.
Tutorials
Serving LLMs at Scale: A Practical Guide with KServe, KEDA,VLLM & Kubernetes
medium.com
This tutorial shows how to serve and scale large language models on Kubernetes with KServe, use KEDA for demand-based scaling, and run inference with vLLM.
Should your daily batch job live inside your main application?
dev.to
This tutorial shows how to move a daily Spring Boot batch job out of the web app and run it as a Kubernetes CronJob so memory spikes do not take down the API.
Policy-Driven Security and Governance in Kubernetes AI Platforms
hackernoon.com
This tutorial teaches how to secure Kubernetes AI platforms with RBAC, NetworkPolicy, secrets, resource limits, Kyverno, OPA Gatekeeper, and automatic policy checks.
5 Kyverno Policies to Secure Your Kubernetes Cluster
devarshishimpi.hashnode.dev
This tutorial shows you how to secure Kubernetes with five Kyverno policies for resource limits, image tags, labels, restricted privileges, and signed container images through YAML examples.
Deploying an OSS Kubernetes security console
cloudsecburrito.com
This tutorial deploys that same open-source Kubernetes security console with Helm and then checks that vulnerability, posture, policy and runtime findings are really flowing, not just that the pods started.
Kubernetes jobs
Infrastructure Architect with Everpure
Salary: $156.1K to $234.5K a year
Location: remote from
Tech stack: Kubernetes, On-Prem, Google Cloud Platform, Kubernetes, VMware, AWS, Azure, Docker, Openstack, CI/CD pipeline applications
Developer Advocate with Nebius
Salary: $108K to $253K a year
Location: remote from
Tech stack: Kubernetes, Kubernetes, AWS, Azure, Docker, GCP, Git, Vertex AI, Sglang, LangSmith
Site Reliability Engineer with Everpure
Salary: €4.05K to €59.4K a year
Location: based in the office in Prague, CZ
Tech stack: Kubernetes, Kubernetes, Public Cloud, Linux, alerting, alerts, metrics, logs, observability, GitOps
Software Engineer with Sopra Steria
Salary: $1.17L to $3.06L a year
Location: based in the office in Noida, IN
Tech stack: Kubernetes, Kubernetes, OpenShift, Elastic Stack, Kibana, Elasticsearch, SonarQube, Fortify, Beats, Logstash
Software Engineer with Sopra Steria
Salary: $1.17L to $3.06L a year
Location: based in the office in Bengaluru, IN
Tech stack: Kubernetes, Kubernetes, OpenShift, Elastic Stack, Kibana, Elasticsearch, SonarQube, Fortify, Beats, Logstash
Discover more Kubernetes jobs on Kube Careers →
Code & tools
KubeIntellect: Human-Governed AI for Kubernetes
github.com/MSKazemi
KubeIntellect lets you troubleshoot Kubernetes in plain English by checking kubectl, Prometheus, and Loki, while requiring human approval before it changes the cluster.
kubectl-rsync: Sync Files with Kubernetes Pods
github.com/major1201
kubectl-rsync is a kubectl plugin that uses rsync to copy only changed files between your computer and Kubernetes pods, with compression, progress tracking, and resumable transfers.
Kagent: Kubernetes-Native AI Agent Framework
github.com/kagent-dev
Kagent is a Kubernetes-native framework for building and running AI agents as custom resources, with multiple model providers, reusable MCP tools, a UI, CLI, and tracing.
HelmSharp: In-Process Helm Workflows for .NET
github.com/GaTTGeng
HelmSharp is a managed .NET library for rendering Helm charts and managing Kubernetes releases in-process, including packaging, repositories, upgrades, rollbacks, and storage, without the Helm executable.
obleth: Multi-Tenant AI Gateway for Shared GPUs
github.com/thediymaker
obleth is a multi-tenant AI gateway that fairly shares GPU capacity, routes OpenAI-compatible requests to the right model, starts Slurm jobs, and tracks cost and energy use.
Other interesting projects:
Upcoming Kubernetes events
Oct
15
Online conference organized by CNCF.
This is a virtual event
This is a free event.
Oct
16
In-person conference organized by Sidero Labs.
Location: Amsterdam, NL
This event requires an entrance fee
Oct
18
In-person conference organized by OmniOpenCon.
Location: Bucharest, RO
This event requires an entrance fee
Oct
19
Kubernetes Community Days UK - Edinburgh 2026
In-person conference organized by KCD UK.
Location: London, GB
This event requires an entrance fee
Oct
20
In-person conference organized by All Things Open.
Location: Raleigh, NC, USA
This event requires an entrance fee
Oct
22
Online workshop organized by LearnKube.
This is a virtual event
This event requires an entrance fee
Kubernetes call for papers
5
days
The Call For Paper is open until 19 October 2026 at UTC. More info →
Location: Los Angeles, CA, USA
In-person conference organized by SCaLE.
The conference starts on the 1 April 2027.
- Apply here
40
days
The Call For Paper is open until 23 November 2026 at UTC. More info →
Location: Vilnius, LT
In-person conference organized by DATA MINER.
The conference starts on the 27 November 2026.
- Apply here
5
days
Cloud Native AI + Inference Day Europe
The Call For Paper is open until 19 October 2026 at UTC. More info →
Location: Barcelona, ES
In-person conference organized by CNCF.
The conference starts on the 15 March 2027.
- Apply here
18
days
The Call For Paper is open until 1 November 2026 at UTC. More info →
Location: Los Angeles, CA, USA
In-person conference organized by DevopsDays.
The conference starts on the 2 April 2027.
- Apply here
1
days
Global Summit on Data Science and Cloud Computing
The Call For Paper is open until 15 October 2026 at UTC. More info →
Location: Rome, IT
In-person conference organized by Noveltics Group.
The conference starts on the 20 October 2026.
- Apply here
59
days
The Call For Paper is open until 12 December 2026 at UTC. More info →
Location: Apenheul, NL
In-person conference organized by Stichting Meetups.
The conference starts on the 11 December 2026.
- Apply here
22
days
Open Source Conference Kolkata
The Call For Paper is open until 5 November 2026 at UTC. More info →
Location: Kolkata, IN
In-person conference organized by Open Source Con India.
The conference starts on the 5 December 2026.
- Apply here
1
days
GENAIX: Global Generative & Agentic AI Summit
The Call For Paper is open until 15 October 2026 at UTC. More info →
Location: Singapore, SG
In-person conference organized by The People Events.
The conference starts on the 16 April 2027.
- Apply here
79
days
The Call For Paper is open until 1 January 2027 at UTC. More info →
Location: Raleigh, NC, USA
In-person conference organized by Devopsdays.
The conference starts on the 27 April 2027.
- Apply here
Until next time!
— Gulcan