Process & Resource Monitoring (Top/Htop/Ps/Systemd)

Stop guessing why your system is thrashing when load average lies to you

🧩 The Challenge

You look at the load average, see a 12.0, and assume the sky is falling. But then you check the CPU usage and it’s practically asleep, which makes you want to throw your monitor out the window because the server is still crawling.

💡 The Fix

Stop staring at the load average and start looking at PSI (Pressure Stall Information) metrics to see if your tasks are actually waiting on CPU, memory, or I/O resources. It’s the only way to know if you’re actually bottlenecked.

cat /proc/pressure/cpu /proc/pressure/io /proc/pressure/memory

⚙️ Why It Works

These files output the percentage of time tasks were delayed by resource pressure over different windows, cutting through the noise that traditional load averages create. It’s essentially the truth serum for why your app feels slow when the CPU doesn’t show it.

🚀 Pro-Tip: If the ‘some’ column is spiking in /proc/pressure/io, go buy faster disks instead of chasing down rogue processes.

Linux Tips & Tricks | © ngelinux.com | 9/2/2026