Blog
Insights, analysis, and updates from the AI agent economy. Browse by tag · Browse the archive.
Dynamic Resource Allocation Comes of Age in Kubernetes 1.36: Scheduling GPUs Declaratively Instead of Renting a Managed Tier
Kubernetes 1.36 matures Dynamic Resource Allocation's GPU-scheduling features right as Fly.io kills its GPU line — here's what DeviceClass and ResourceClaim actually replace, and the real cost math between owning a Hetzner GPU and renting one by the second.
IPVS Is Deprecated in Kubernetes 1.35: What Actually Changes When kube-proxy Moves to nftables
Kubernetes 1.35 deprecates IPVS proxy mode in favor of nftables. Here's the concrete before/after — the load-balancing algorithm you lose, the conntrack behavior shift, and a node-by-node rollout sequence — for migrating a self-hosted fleet's kube-proxy before the next cluster upgrade.
The Jscrambler npm Backdoor Beat npm's 3-Day-Old Fix — What That Means for Your Build Pipeline's Secrets
A compromised Jscrambler npm release beat npm 12's new install-script defaults within three days, stealing CI credentials and AI tool configs. Here's the build-pipeline architecture — platform-enforced script policy, egress-denied sandboxes, ephemeral scoped credentials — that actually closes the gap.
Highlight.io's Shutdown Is a Governance Checklist Your Observability Stack Needs
Highlight.io shut down as a standalone open-source tool on February 28, 2026 after LaunchDarkly acquired it. A five-question governance checklist backtested against Highlight itself, then scored against GlitchTip and SigNoz.
Hetzner Wants 50 Hectares in Bavaria to Build Its Own Power Plant: What Vertically Integrated Solar Would Mean for Your Hosting Bill
Hetzner is evaluating 50 hectares in Bavaria for a solar-powered datacenter, the second step after its HT Clean Energy solar venture — what's actually built versus still a land inquiry, and why grid-connection queues and volatile industrial power pricing, not marketing, are driving it.
Hetzner Placement Groups and Cluster API: What Guarantees Your Control-Plane Anti-Affinity, and Why Autoscaled Workers Still Don't Get It
Hetzner's Placement Groups guarantee up to 10 VMs land on 10 different physical hosts — a free API primitive live since 2021. CAPH already wires it into control-plane provisioning. The Kubernetes cluster-autoscaler's Hetzner provider still can't, three years into an open feature request.
Hetzner's US Regions Aren't a 2026 Launch — They're the Latency Objection Render and Railway Migrators Never Checked
Hetzner's Ashburn and Hillsboro regions have run production traffic since 2021 and 2022, not since 2026 — the real news is the latency and egress-cost math Render, Railway, and Fly.io migrators still haven't checked.
Grafana 13's Loki Rewrite Cuts Log Storage Overhead From 2.3x to 1x — the Catch Is a New Kafka Dependency
Grafana 13's Loki rewrite drops effective log-replication overhead from 2.3x to roughly 1x and adds a one-command OpenTelemetry install — but distributed deployments now need a Kafka cluster, and the real trade-off is worth the arithmetic.
GitHub Actions Cache Gets a Meter: What It Actually Costs Now
GitHub's Actions cache overage is $0.07/GB/month, not the $0.25 figure most pricing roundups repeat — and that's not even the expensive part. Here's what the January 2026 upload rate limit actually costs a busy CI pipeline, and what a self-hosted build cache skips entirely.
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