Blog
Insights, analysis, and updates from the AI agent economy. Browse by tag · Browse the archive.
KYAML Goes Stable in Kubernetes v1.37: What a Canonical YAML Subset Means for the Manifests a Git-Push PaaS Generates on Every Deploy
Kubernetes 1.37 promotes KYAML — a strict, unambiguous YAML subset with a stable kubectl output — to GA. What canonical manifests change for platforms that generate YAML on every deploy, and why compliance belongs at emit time.
'Kubernetes Is the PaaS': What the Portability Thesis Gets Right, and the Operations Bill It Skips
Hacker News keeps rediscovering the thesis that Kubernetes is a vendor-neutral PaaS layer. The portability half holds up — workload manifests really do run anywhere — but the operations half never disappeared: upgrades, etcd, networking, and node lifecycle still need an owner.
Kubernetes 1.37 Takes HPA Scale-to-Zero to Beta: What Idle Finally Costs vs Fly.io, Render, and Railway
HPA scale-to-zero is beta and on by default in Kubernetes 1.37. A duty-cycle idle-bill table pits native minReplicas: 0 against Fly.io, Render, and Railway — plus the HTTP buffering gap to close before idle costs zero.
Kubernetes Enters Its Boring Era: What Three Stability-First Releases Buy a Part-Time Platform Team
Kubernetes 1.36 through 1.38 are deliberately boring releases — steady graduations, one deprecation at a time. A ledger of what that saves a small Cluster API fleet during upgrade season, and the patch treadmill it doesn't cancel.
Five Surprising Ingress-NGINX Behaviors to Audit Before Migrating to Gateway API
Ingress-NGINX retired in March 2026, and a clean converter run is not a correct migration. Audit these five behaviors — regex quirks, vanishing redirects, and untranslatable annotations — before flipping traffic to Gateway API routes.
HPA Scale-to-Zero Is Beta in Kubernetes 1.37: The Two Deadlocks That Strand Workloads at Zero
Kubernetes 1.37 made HPA scale-to-zero a default-on beta — but a metric rule that can never fire and a status condition that was never set can strand workloads at zero replicas forever. A pre-flight checklist for platforms: safe wake-up metrics, zero-lock detection, and single ownership of the replica count.
Hetzner's €5.99 Server Is a Number You Can't Buy: What the Missing Cheap Tier Does to the Self-Hosting Price Anchor
Hetzner's €5.99 cloud server is still on the price list but every Cost-Optimized model is marked unavailable, forcing new deployments onto tiers costing 3.34× to 4.24× more. This post traces the two 2026 price hikes, the DRAM shortage behind them, and how to re-anchor self-hosting budgets to the catalog you can actually order from.
Heroku, Seven Months Into Sustaining Mode: Where the Leavers Went, and the Fourth Option the Guides Never Priced
Seven months after Heroku's sustaining-mode announcement, the 2026 guides route every leaver to Render, Railway, or Fly.io. We audit four guides' routing consensus, triangulate where teams actually landed, and price the row they all skip: the same dyno-hours as utilization on one flat box.
The Heroku Exit Playbook's Missing Half: Pipelines and Review Apps on Machines You Own
The Heroku exit playbook moved apps, add-ons, and data — but release process stayed behind. A concrete mapping of Pipelines promotion and Review Apps onto Kargo, GitHub Actions environments, and per-PR namespaces, with the honest cost of replicating automatic per-PR databases.
Subscribe
New posts land in your reader as soon as they publish. Pick a format — all three carry the same posts.
Current feeds keep roughly two days of posts so daily polling does not miss a burst. Older entries stay reachable from the feed's next-page link in readers that follow it, or from the blog archive.
Following one topic instead? Browse tags