We’ve successfully put MariaDB on a strict diet, but don’t pop the champagne just yet. In the cramped 512MB “apartment” of a $5 VPS, Nginx, PHP-FPM, and MariaDB have to live together. If one gets greedy, the OOM (Out of Memory) Killer will show up and start swinging its scythe. Today, we’re going to tighten … 더 읽기
TL;DR: If your MariaDB service keeps stopping on low-memory cloud instances (like AWS Lightsail $5 plan), it’s likely due to the OOM (Out of Memory) Killer. The simplest fix is adding a 2GB swap file to provide a safety net for your RAM. 1. The Problem: Memory Starvation on 512MB Instances When you launch a … 더 읽기
(For the Korean version, click here) If your internal processes are running fine and your logs are clean, but users still can’t access your services, it’s time to look outside. Whether it’s “Why can’t clients connect to my ready server?” or “Why can’t our server talk to that external API?”, these are questions we must answer. … 더 읽기
(For the Korean version, click here) As an engineer, you will eventually face critical moments when a service suddenly slows down or becomes unresponsive. While Parts 9 and 10 focused on tracing the past through logs, today we focus on the present: “Who is eating up my server’s resources right now?” In this post, we will … 더 읽기
(For the Korean version, click here) For an engineer, logs are like “SOS signals sent by the server.” If we learned how to find the right files in Part 9, it is now time to read and interpret the messages pouring out of them in real-time. Today, we will cover real-time log monitoring techniques—how to pinpoint … 더 읽기
(For the Korean version, click here) For an engineer, digging through logs is a daily occurrence. Whether you are troubleshooting an outage or tracing system flow, logs are the first place you look. In a sea of tens of thousands of files, how quickly you find the data you need defines your skill level—and is the … 더 읽기
(For the Korean version, click here) Creating is easy. However, true expertise is revealed during ‘deletion.’ In virtualization or cloud environments, you often encounter a “Busy” or “Cannot be removed” message when trying to delete a volume. This happens because the OS still recognizes and is actively using that volume. A truly ‘clean deletion’ doesn’t just … 더 읽기
(For the Korean version, click here) In our last post, we covered the basics of LVM. Today, we’re diving into the “killer feature” of LVM: Online Expansion. If you’re a sysadmin, you’ve probably heard this before: “The /data partition is at 90%! We need more space ASAP!” In the old days, this was a total … 더 읽기
(For the Korean version, click here) Now that we’ve got Rocky Linux running from our last post, let’s give it some more breathing room. We’re going to add a new hard drive and set up a proper storage area. You might ask, “Why not just use a simple partition?” Honestly, in a real production environment, … 더 읽기
(For the Korean version, click here) We live in an era where you can spin up a server with just a single click—a “one-click setup”—on cloud platforms like AWS or GCP. It’s incredibly convenient. So, it’s fair to ask: “Does an engineer really need to know how to install an OS manually anymore?” In my … 더 읽기