Dora Noda
Software Engineer
2319 posts · View all authors
Dokploy Is Apache 2.0 Again — But SSO, Audit Logs, and RBAC Now Live Behind a New Source-Available License
Dokploy's January 2026 license split moved SSO, audit logs, RBAC, and white-labeling behind a new source-available license. We pulled the actual GitHub tree to see exactly what changed and what didn't.
Encore.dev Promises No Terraform — Until You Self-Host
Encore.dev's pitch is that declaring a database in your app code replaces writing Terraform. That's true — on Encore Cloud. Self-host the same app and you're back to provisioning infrastructure by hand and handing Encore a config file. A worked look at where the automation actually stops, and what it costs when it doesn't.
Preventing etcd Zombies on Upgrade to 3.6: A Cluster API Operator's Pre-Flight Checklist
Kubeadm still pins etcd to v3.5.24-0 for Kubernetes 1.31 through 1.33 — one patch below the v3.5.26 floor etcd's maintainers say you need before v3.6. Here's why Cluster API's machine-replacement upgrade model skips that safety margin entirely, and the pre-flight checklist to run before you cross the boundary.
Firebase Studio Is Shutting Down. Here's What 'Export My Code' Actually Means Across Every Vibe-Coding Platform
Firebase Studio dies for good on March 22, 2027. Its own migration docs reveal exactly what exports cleanly and what doesn't — and the same checklist exposes how much of Bolt.new, Lovable, Replit, and Base44's output is actually yours to keep.
Fly.io's Managed Postgres Costs $38/Month Before You've Added the App
Fly.io's Managed Postgres starts at $38/month plus $0.28/GB storage — and that's before the app talking to it is billed. A line-by-line worked comparison against running Postgres on the same Hetzner box already hosting the app.
GitHub Actions' Self-Hosted Runners Went Dark for Five Hours: The Outage That Proves 'Self-Hosted' Isn't 'Self-Owned'
A TLS certificate expired on a GitHub endpoint on July 19, 2026, and every self-hosted GitHub Actions runner on the internet went idle for five hours — even though the machines themselves never stopped running. Here's what actually broke, and why owning the compute was never the same as owning the pipeline.
GitHub Agent HQ Runs Claude, Codex, and Copilot on the Same Repo — Here's the Trust Boundary Your MCP Server Actually Needs
GitHub Agent HQ runs Claude, Codex, and Copilot side by side on the same repo under one identity model — here's why a deploy platform's MCP server should authenticate the calling app, not the AI vendor behind it.
GKE Labs' OpenRL Runs Fine-Tuning as a Kubernetes Workload: What Owning the Loop Actually Saves Over a Managed API
Google's GKE Labs open-sourced OpenRL, a self-hosted, Tinker-compatible fine-tuning API for Kubernetes. Here's the worked cost math on what running your own LoRA fine-tuning loop actually saves over Together AI's managed API — and what it doesn't.
GPU Inference's Hidden Egress Tax: The 22-31% Line Item Nobody Prices In
2026 cost teardowns keep finding the same thing: GPU-hours are barely half an inference bill once a service actually serves traffic at volume. Here's the worked math on AWS, Azure, and GCP showing where the other 22-31% goes — and an honest look at what self-hosting the serving layer removes versus what it doesn't.