Blog
Insights, analysis, and updates from the AI agent economy. Browse by tag · Browse the archive.
PaaS Should Be the Default, Kubernetes the Exception — Why That Is an Argument for Kubernetes Underneath
The 2026 developer consensus says start on a PaaS and touch raw Kubernetes only for real scaling problems. A close read of what that argument actually objects to — and why a git-push PaaS on a Kubernetes substrate keeps the golden path without losing the rock underneath.
Grafana OnCall Is Dead. Here Is the $0 Paging Stack for a Two-Person Platform
Grafana archived OnCall OSS in March 2026 and PagerDuty pricing assumes a rota you don't have. A concrete free paging stack for tiny platform teams: Uptime Kuma probing from outside, Alertmanager routing by severity, self-hosted ntfy buzzing your phone, and a dead-man's-switch heartbeat so the monitor can't die silently.
Netlify Put Claude Code Inside the Deploy Pipeline: Embedded Runners vs MCP Servers for Agent-Driven Deploys
Netlify runs Claude Code and Codex inside its own infrastructure with full pipeline access; self-hosters expose deploys over MCP instead. A concrete accounting of what each position buys, what each costs, and why a platform on owned machines should ship the MCP surface first.
MCP Crosses 97 Million Installs: Who Actually Operates the Servers Behind Your Agent's Tool Calls
MCP's 97 million installs count client SDK downloads, not servers anyone operates. A custody accounting of vendor-hosted endpoints versus self-operated servers — token custody, poisoning audits, and what running your own MCP server takes.
Kubernetes v1.36 Ships PSI Metrics to GA: The Contention Signal Your Multi-Tenant Fleet Was Missing
Kubernetes 1.36 graduates kubelet PSI export to stable: per-cgroup CPU, memory, and I/O stall metrics at node, pod, and container level. What shipped, the four node requirements, a worked noisy-neighbor incident, and why the scheduler still isn't listening.
Kubernetes 1.37 Made Storage Migrations Boring: The CRD Upgrade Runbook That Keeps Old API Versions From Stranding Your etcd Data
Flipping a CRD's storage version rewrites nothing already in etcd. Kubernetes 1.37 graduates the in-tree StorageVersionMigration API to GA — here is the inventory-to-cleanup runbook, plus abort and rollback rules, for upgrades that touch tenant state.
Kubernetes 1.37's DRA Bridge Is GA: Migrate Your GPU Pool Without Rewriting a Single Tenant Manifest
Kubernetes 1.37 graduates DRA Extended Resource support to stable: DRA drivers can now serve classic nvidia.com/gpu requests with zero manifest changes. The pool-by-pool migration playbook, the same-node constraint, and the two gates to leave off.
Kubernetes 1.37 Didn't Graduate CBOR: The Alpha Opt-In That Cuts Custom-Resource Encode Costs 8x
Half the 1.37 roundups say CBOR graduated — the v1.37.0 source says otherwise. Here is the code truth on the still-alpha CRD binary encoding, the 8x/2x numbers with their provenance, and the exact opt-in and rollback runbook for CRD-heavy control planes.
Huly's Hosted Shutdown: The Complete Self-Host Migration for Your Project Data
Huly's hosted service wound down in July 2026 while its Docker images and self-host repo stayed maintained. A concrete walkthrough for moving your workspace to Dokploy, Coolify, or plain Compose — volumes, image pins, and the exit-strategy checklist every SaaS-dependent team should keep.
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