Krzysztof Rączkiewicz

DevOps Reality Check: When "Simple" Deploys Meet 3-Server Complexity Most tutorials make deployment look like a single click. In a professional production environment, it is rarely that straightforward (unless you are lucky to have Vercel or similar). In this case, unfortunately it wasn’t as simple. I recently worked through a migration involving a 3-server ecosystem where automation was the only way to maintain my sanity: • The Compute Node (.21): The engine room running Nginx and PHP 8.3, now serving the application from a dedicated subfolder. • The Database Node (.22): A separate, secure environment for all data, keeping the application node lean and fast. • The Storage Node (.20): A remote NFS mount for media, ensuring files stay safe even if the application nodes are replaced. Managing this setup required constant coordination between secure VPN access and SSH protocols. Because these servers live behind a strict firewall, a standard cloud runner could not reach them. The solution? I implemented a self-hosted runner. By placing the runner inside the internal network, I could securely bridge the gap between the GitHub repository and the private infrastructure. This allowed me to automate the move of the entire architecture into a new directory while keeping the storage "tunnel" alive. If your automation is not precise, your app loses its connection to its files instantly. The Lesson In a multi-tier architecture, your CI/CD pipeline is the glue holding your infrastructure together. Complexity is fine, as long as you have the automation and the right security bridges to back it up. It is all fun and games until a mount point disappears or a VPN block stops a deploy, but that is why I build robust pipelines! #Development #DevOps #GitHub #API

4

DevOps is dying as a profession, and that is actually a good thing. The original DevOps manifesto was simple: developers and operations must share responsibility. But many businesses took a shortcut. They created "DevOps departments," which became a third silo. Instead of teaching developers about infrastructure, we gave them a new person to throw code at. This created an impossible cognitive load for developers. You cannot expect a feature dev to also be an expert in Kubernetes, Terraform, and security. It doesn’t scale. Platform Engineering is the solution. It treats DevOps practices as a product. Instead of telling developers "you must do this," Platform Engineering says, "here is a Golden Path that does it for you". It allows people to focus on business value instead of fighting tools. AI is the final catalyst. AI now generates thousands of lines of code in seconds. We cannot hire enough DevOps engineers to manually review and deploy this volume. It is economically impossible. We need IdP to act as an "immune system": - Compliance-as-Code: Automatically blocking unsafe AI-generated code. - Self-healing: Platforms that roll back failures without human intervention. - Templates: Giving AI specific patterns so it doesn't "hallucinate" bad infrastructure. DevOps as a job title is dying. But DevOps as a culture is finally becoming real through Platform Engineering and AI. How is your team handling the "AI code explosion"?

364

87 Comments

Welcome to the first 2026 edition of Tidbits of Technology🎉! The meeting was led by Mateusz Smagieł, who explored how authorization architecture evolves from static code to dynamic, runtime evaluation. We learned that operability and resilience must be prioritized over the DRY principle, as removing duplication is not worth risking availability. Check out the graphics below to learn more about the architectural decisions and trade-offs we discussed! #Microservices #SidecarPattern #TidbitsOfTechnology #IT

18

How can you combine faster delivery with high security and smart automation? 🛡️    In our latest article, Principal Systems Engineer Sławomir Bednarczyk and DevOps Engineer Mateusz Kozieł explain how they've integrated AI into a CI/CD pipeline to boost security scanning.    This practical walkthrough shows an intelligent process flow that leverages a retrieval-augmented generation (RAG) AI agent, a vector database with CIS Benchmarks and the Trivy scanner to automatically generate merge requests with security patches. Traditional security scanning often results in information noise. Find out how to shift the burden from reporting to remedying.     👉 Read more here: https://lnkd.in/dvY_RU2A   #DevOps #DevSecOps #AI #Security 

28

1 Comment