Blog - Konrad Kowalski (rootsher)
posts (112)
- Kubernetes from the Inside: Cluster Architecture Basics7/7
Managed, Self-managed, Cloud, On-premises, VM and Bare Metal
Managed, cloud and bare metal are not three alternatives but three separate questions - who manages Kubernetes, where the infrastructure runs and what the…
- CPU in Kubernetes from the kernel up8/12
Why 1 CPU does not mean one core
CPU time and CPU placement are two different mechanisms. A limit says how much you may consume; affinity and cpuset say where a task may execute.
- Kubernetes from the Inside: Cluster Architecture Basics6/7
How Kubernetes Components Work Together
One workload from kubectl apply to a running Pod, step by step through the API Server, etcd, controllers, the scheduler, the kubelet, the runtime and CNI…
- Kubernetes from the Inside: Cluster Architecture Basics5/7
DNS and CoreDNS in Kubernetes
A Pod's IP address is a poor interface for applications, because it changes. CoreDNS turns a stable name into a Service IP, and only then the datapath delivers…
- CPU in Kubernetes from the kernel up7/12
Requests, Limits and QoS: how Kubernetes classifies Pods
A QoS class is the result of how requests and limits are configured, not a separate CPU allocation mechanism. Guaranteed is not a dedicated core.
- Part I: From Prompts to Agentic SDLC8/9
Beyond Dynamic Workflows: when the workflow leaves Claude
The eighth level separates agent reasoning from durable orchestration: retries, webhooks, approvals, queues, long waits and workflow state outside Claude's…
- Kubernetes from the Inside: Cluster Architecture Basics4/7
Networking in Kubernetes
CNI gives Pods their addresses and connectivity, while kube-proxy or an eBPF datapath makes traffic sent to a Service's virtual address reach the right…
- Kubernetes from the Inside: Cluster Architecture Basics3/7
The Worker Node and the Data Plane
The control plane decides, but the worker node executes - the kubelet looks after the Pods assigned to the machine and, through CRI, asks the container runtime…
- CPU in Kubernetes from the kernel up6/12
PSI: how Linux measures real CPU pressure
Utilisation says how much CPU a workload got. PSI measures the time it lost because it wanted to execute and had nothing to execute on.
- Kubernetes from the Inside: Cluster Architecture Basics2/7
The Kubernetes Control Plane
The control plane is not one brain but four components with separate roles - the API Server communicates, etcd stores the state, the scheduler picks a node and…