How to Profile a Slow Server with Perf on Linux

Lucas and Luna dive into perf, the Linux profiling tool, with a real production scenario: a web server experiencing high CPU latency. They walk through capturing performance data using perf record, analyzing hot functions with perf report, and visualizing call stacks with flamegraphs. You'll learn how to pinpoint the exact function consuming CPU cycles—whether it's a Lua script in Nginx or a kernel driver. The episode covers sampling frequency, call-graph modes, hardware counters like instructions per cycle, and syscall tracing with perf trace. By the end, you'll have a repeatable workflow for diagnosing CPU bottlenecks, applicable to any Linux server running kernels 5.x and above. This is Episode 130 of Linux Server Admin with Fexingo, and includes a brief behind-the-scenes on how listener support keeps the show ad-free. Perfect for sysadmins, DevOps engineers, and anyone who manages production servers and wants to move beyond guesswork. #LinuxServerAdmin #FexingoBusiness #TechnologyPodcast #Linux #Perf #CPUProfiling #PerformanceMonitoring #Flamegraph #ServerOptimization #KernelTuning #Sysadmin #DevOps #Nginx #ProfilingTools #SystemAnalysis #PerformanceDiagnostics #LinuxPerformance #Episode130 Keep every episode free: buymeacoffee.com/fexingo