513 posts tagged with "Kubernetes"
Container orchestration, Cluster API, and self-hosted control planes
One Script, Five Hours, Three Dead Layers: What Azure's West US Outage Reveals About the Shared Fate Inside Every Managed Region
On July 23, 2026, a maintenance-automation bug deleted network routes across Azure West US for five hours, taking down AKS, managed Postgres, and ingress together. Why zone redundancy couldn't help, what multi-region survival actually costs, and what owned hardware decouples.
Buildpacks RFC 331: Getting the Extend Phase Off Vendored Kaniko
A Dockerfile hides inside every Dockerfile-free buildpack build — applied by a kaniko copy baked into the lifecycle image. RFC 331 would pull the extender out behind a configurable URI; Chainguard's kaniko fork bought time but didn't fix the coupling. The before/after table and the six-item operator checklist.
containerd 2.1 Is Dead and the CVE Has No Patch: Rolling a Forced Runtime Upgrade Across a Cluster API Fleet
containerd 2.1 went end-of-life with no patch for Critical CVE-2026-46680, a runAsNonRoot bypass. Which versions are safe, how to pick a landing zone, and the five-step Cluster API runbook: audit, image, template, roll, verify.
Coolify v5's Multi-Server Bet: What the Most Popular Single-Box PaaS Hitting Its Ceiling Teaches About Cluster Orchestration
Coolify named multi-server scalability as v5's core feature — with no release date. A concrete accounting of what a second server buys you in v4 today, what v5 is actually building, and why the gap is a scheduler-shaped hole only cluster orchestration fills.
Borrow the Claim, Skip the Cloud Sprawl: What Crossplane's Modelplane Teaches a Self-Hosted PaaS About GPU APIs
Modelplane puts GPU inference fleets behind a Crossplane claim API so teams request an endpoint, not a machine. The before/after YAML, a six-row borrow-vs-skip table for a Hetzner CAPI fleet, and the €184/mo card math that decides it.
CPU Lies: What a Git-Push PaaS Should Autoscale Tenants On Instead
CPU sits at 35% while your service melts — because it was waiting, not working. A concrete pipeline for autoscaling tenants on in-flight requests and queue depth, the adapter-vs-KEDA decision, multi-tenant guardrails, and what Render and Railway get right.
Do I Really Need Kubernetes? What the PaaS-First Backlash Gets Wrong
HN threads say a PaaS is the default and Kubernetes is overkill. The honest answer separates operating a cluster by hand from owning a git-push platform built on it — plus the concrete second-machine test that decides which side you are on.
etcd 3.7 Boots From the v3 Store: What Changes About Backups and Recovery on a Self-Hosted Fleet
etcd 3.7 boots entirely from the v3 store, with the last v2 leftovers slated for removal in 3.8 — what that means for snapshots, compaction, and corruption recovery, plus a six-step fleet-wide backup drill to run before your next Kubernetes upgrade.
No More Annotations: Gateway API v1.5 and v1.6 Make PaaS Ingress 100% Portable YAML
Gateway API v1.5 and v1.6 moved the CORS filter, TLSRoute, and TCP/UDP routes into the Standard channel, so a git-push PaaS can express a tenant's whole ingress layer as portable YAML with zero annotations. Includes the GRPCRoute timeline correction and the four-item conformance checklist to run before tenants rely on it.