Tag: #kubernetes
Writing on GPUs, LLMs, MLOps, Kubernetes — and mindset · 249 posts
249 posts are tagged #kubernetes; the most recent was published on 2026-09-10.
Often tagged together: #devops 62 #2026-03 59 #ingress 32 #networking 32 #security 27
Most read in the last 90 days:
CloudNativePG deep dive — who looks after PostgreSQL inside Kubernetes
Why it does not use a StatefulSet, why PID 1 in the pod is not postgres, what the three services do, and why a base backup and WAL must be combined before anything can be restored. The features and role of CNPG, explaine
2026-09-10 · 7 min read #kubernetes#postgresql#cloudnativepg#backup#operatorScanning the production image with Trivy — 768 findings, and the real work was three packages and one Dockerfile line
From a signature-verification bypass in the login library (CVE-2026-27962) to a digest-pinned base image that could never receive OS security fixes. The scanner counting itself, false positives in secret detection, and a
2026-09-10 · 6 min read #security#cve#trivy#docker#pythonA Three-Line Config File Killed Four GPUs for a Week — How containerd Drop-in Merging Really Works ♪ Listenable
The nodes stopped advertising GPUs. The nvidia runtime was written out cleanly in the config file, but it was nowhere to be found in containerd config dump. The culprit was a three-line registry config I had added a week
2026-08-26 · 12 min read #kubernetes#containerd#gpu#nvidia#troubleshootingBuilding a Hands-On Learning Platform on a Home Server — Putting Kubernetes Inside an Unprivileged Pod
A write-up of putting a learning platform on a 7-node Kubernetes cluster at home: a learner presses a button, a dedicated container comes up, a browser terminal attaches to a real Linux shell, and at every step the serve
2026-08-20 · 16 min read #kubernetes#cilium#security#homelab#kwokLabHub — I opened a hands-on platform where you learn on a real server, in the browser
Some things do not stick from reading. There is no substitute for attaching to a server, typing a command, getting it wrong, and finding out why. LabHub is a learning platform that puts that loop inside the browser. Pres
2026-08-20 · 8 min read #labhub#kubernetes#devops#education#hands-onSelf-Hosting Langfuse — Deployment Paths, Secrets, and What Catches You on First Boot
Running Langfuse yourself means standing up two containers and four datastores at once. Working from the official documentation, this post lays out the docker compose path and the Helm chart path, which secrets you have
2026-08-14 · 9 min read #observability#langfuse#self-hosting#docker-compose#kubernetesOpen Source Worth Watching Right Now (3) Infrastructure and Databases
Databases and infrastructure are the field where license changes and forks redrew the board. From analytics engines and embedded databases to Postgres extensions, Kubernetes operators, and IaC, this post introduces 11 op
2026-08-12 · 5 min read #open-source#database#infrastructure#postgresql#kubernetesThe Generational Shift in Container Infrastructure — What Eleven Projects Left Behind
Eleven projects that were once standard parts of a container infrastructure and have since been replaced, documented using only official notices and repository archive status as evidence. rkt, dockershim, Classic Swarm,
2026-08-12 · 10 min read #open-source#kubernetes#container#docker#infrastructurevLLM Metrics — What to Chart and What to Alert On
The series vLLM exposes answer questions GPU metrics cannot: how many requests are running versus waiting right now, how full the KV cache is, how long until the first token. This post reads the official vLLM documentati
2026-08-12 · 7 min read #gpu#kubernetes#vllm#prometheus#observabilityMIG and Time-Slicing — Two Ways to Share One GPU
There are broadly two ways to put multiple workloads on one GPU: time-slicing, which divides time, and MIG, which divides hardware. Despite sounding similar, their isolation guarantees are nothing alike. This post works
2026-08-12 · 6 min read #gpu#kubernetes#mig#time-slicing#nvidiaDCGM Exporter — GPU Utilization Is Not What You Think It Is
DCGM Exporter is the standard path for exposing GPU telemetry in Prometheus format, but the utilization-style metric that ends up on nearly every dashboard does not measure what people expect. This post reads the default
2026-08-12 · 9 min read #gpu#kubernetes#dcgm#prometheus#observabilityA GPU Troubleshooting Playbook — Fix the Layers, Then Walk Down
The biggest waste in diagnosing GPU problems on Kubernetes is poking around without an order. A pod that will not schedule, a pod that runs but cannot see the GPU, a driver and toolkit version mismatch, memory exhaustion
2026-08-12 · 7 min read #gpu#kubernetes#troubleshooting#nvidia#gpu-operatorNVIDIA GPU Operator — The Six Pieces You Used to Install by Hand
Running GPUs on Kubernetes used to mean matching six pieces on every node by hand: the driver, the NVIDIA Container Toolkit, the device plugin, DCGM, GPU Feature Discovery, and Node Feature Discovery. The NVIDIA GPU Oper
2026-08-12 · 6 min read #gpu#kubernetes#gpu-operator#nvidia#dcgmGPU Serving SLOs and Alert Design — What to Promise and What to Page On
Putting an SLO on a GPU inference service starts with deciding which metric represents the user experience. Time to first token and throughput eat each other, so setting a target on only one guarantees the other collapse
2026-08-12 · 7 min read #gpu#kubernetes#slo#alerting#prometheusDevice Plugins and GPU Scheduling — Where nvidia.com/gpu Comes From
Kubernetes does not know what a GPU is. What makes a node advertise GPUs as a resource is a device plugin registered with the kubelet, and the name that falls out of it is the extended resource nvidia.com/gpu. This post
2026-08-12 · 6 min read #gpu#kubernetes#device-plugin#scheduling#nvidiaChoosing an On-Prem Lightweight Kubernetes — Splitting k3s, k0s, and RKE2 by Datastore and Regulation
Instead of listing feature checkboxes, this post compares k3s, k0s, and RKE2 only on the axes that actually decide the choice. The first axis is the datastore. Start with SQLite and you can't add server nodes; embedded e
2026-07-31 · 17 min read #kubernetes#k3s#k0s#rke2#on-premisek0s Air-Gapped Install and k0sctl Automation — From Building the Image Bundle to Multi-Node Upgrades
This post lays out, command by command, how to stand up and operate a k0s cluster in an air-gapped network with no internet path. It covers three ways to build an image bundle with k0s airgap list-images and bundle-artif
2026-07-31 · 15 min read #kubernetes#k0s#k0sctl#air-gap#on-premiseThe Complete Guide to Installing k3s in an Air-Gapped Network — From Image Tarball Transfer to a Private Registry and Agent Joins ♪ Listenable
Covers the whole process of standing up a k3s cluster on a network with no internet path at all, command by command. It walks through downloading the airgap image tarball and the k3s binary on a connected machine, buildi
2026-07-31 · 20 min read #kubernetes#k3s#air-gap#on-premise#containerdAir-Gapped Kubernetes Day 2 Operations — How to Stop a Cluster From Dying of Certificate Expiry ♪ Listenable
What actually kills an air-gapped cluster is usually not some spectacular outage but certificate expiry. Let a year go by with nobody touching it and the API server quietly stops, and with no internet you cannot search y
2026-07-31 · 20 min read #kubernetes#air-gap#day2-operations#etcd#certificateDesigning an Air-Gapped Image Import Pipeline — skopeo, Harbor, and a Reimport Runbook That Doesn't Rot
This post designs the pipeline for a problem almost nobody in air-gapped Kubernetes actually gets right: repeatedly and safely bringing container images in. It covers managing the import list as code, the skopeo sync com
2026-07-31 · 17 min read #kubernetes#air-gap#harbor#skopeo#supply-chain