How to Check CPU Usage in Linux

When a Linux server suddenly becomes sluggish—web pages take seconds to load, SSH connections lag, and database queries timeout—the most common culprit is CPU exhaustion. A single rogue script stuck in an infinite loop, or a sudden spike in legitimate web traffic, can consume 100% of your processor’s capacity, leaving no resources for other essential tasks. As a system administrator, your first diagnostic step must always be identifying exactly which processes are eating your CPU cycles.

Understanding CPU Usage in Linux

Unlike a simple speedometer, Linux CPU usage is a complex metric. If you have a quad-core processor, your total available CPU capacity is technically 400%. A single application might consume 100% (maxing out one core) while leaving the other three cores perfectly idle, meaning your server is still healthy overall. Conversely, if a multi-threaded application consumes 390%, your entire system will grind to a halt. Diagnostic tools help you visualize this distribution.

Method 1: The ‘top’ Command (The Universal Standard)

The top command is installed on virtually every Linux distribution by default. It provides a live, continuously updating dashboard of system performance.

1. Open your terminal or connect via SSH.
2. Type top and press Enter.
3. The screen will split into two sections. The top section shows system-wide averages, while the bottom section lists individual processes.

Reading the top output:

• Look at the third line, labeled %Cpu(s). The most important metric here is id (idle). If the id number is low (e.g., 5.0 id), it means your CPU is 95% busy.
• Now look at the process list below. By default, top sorts the list by the %CPU column. The application at the very top of the list is the one consuming the most processing power.
• Note the PID (Process ID) of the offending application. If the process is a frozen script you need to terminate, you will use this PID with the kill command.
• To exit the dashboard and return to the normal terminal prompt, press the q key.

Method 2: The ‘htop’ Command (The Better Alternative)

While top is powerful, it is visually dense and somewhat difficult to read. The htop command is a modernized, interactive alternative that uses color-coding and graphical bar charts to make CPU analysis instantly intuitive.

1. You likely need to install it first.
   • Ubuntu/Debian: sudo apt install htop
   • CentOS/RHEL: sudo dnf install htop
2. Once installed, type htop and press Enter.

Why htop is superior:

• At the very top of the screen, htop displays a visual bar chart for every individual CPU core (numbered 1, 2, 3, 4, etc.). You can instantly see if the load is balanced across the entire processor or if a single core is maxed out.
• The process list is color-coded.
• You can use your mouse (or arrow keys) to scroll vertically through the entire process list and horizontally to read long command strings.
• You can kill a process directly from the interface. Simply highlight the rogue process with the arrow keys and press the F9 key, then press Enter to send the kill signal.
• Press F10 (or q) to exit.

Method 3: Checking Historical Usage with ‘sar’

top and htop only show what is happening right now. If your server crashed at 3:00 AM and you want to know what the CPU usage was at that exact time, you need a historical logging tool like sysstat (which includes the sar command).

1. Install it: sudo apt install sysstat (You may need to enable data collection in /etc/default/sysstat).
2. To view CPU usage for the current day, broken down in 10-minute intervals, simply type:
   sar -u

By utilizing top for quick checks, htop for interactive management, and sar for historical forensics, you can ensure your Linux server’s processing power is always optimized.