Granite blog

Three vendors, three bills, and a server you still patch yourself. Here is what managed services actually take off your plate, and how to consolidate an MVP onto one platform.

Mutable latest tags, invalid branch names, race conditions between runs and rebuilt prod images. Seven docker tags mistakes in GitHub Actions and the fix for each.

You have a working docker-compose.yml and six weeks to launch. Here is a timeboxed, two-hour plan for moving it onto managed cloud services and off your pager.

Your GitHub deploy pipeline is probably fine. The server at the end of it is what wakes you at 2 AM. Here is how to harden both, and when to stop owning the server.

The best cloud hosting for an MVP isn't the cheapest one. It's the one whose bill you can predict. Here's how to compare managed cloud servers on cost certainty.

Go beyond a bare docker ps: filter by status, health, label, and port, build custom tables and JSON output, and use it to debug crashing containers.

A fair, solo-founder look at Railway and Granite.so: what drives deploy speed, how the CLI and Git workflows differ, and what a small SaaS really pays each month.

Your S3 storage bill is rarely about gigabytes stored. Here is where the money actually goes, and a step-by-step plan to move your files somewhere predictable.

Most automatic deployment guides optimize for speed. This one optimizes for predictability: a Docker-based pipeline you can trust, set up in under two hours.

A stack trace tells you where your app crashed, not how it got there. Here is how to configure Sentry breadcrumbs so every error arrives with its own timeline.