Taming CPU Usage Spikes with `systemd` and `ps` in My Home Server Setup

 Posted on 17 May 2026  |  4 minutes  |  696 words  |  Tom

Introduction to CPU Usage Spikes

I’ve had my fair share of CPU usage spikes on my home server, and I’ve learned that they can be caused by a variety of factors, including resource-intensive applications, misconfigured services, or even malware. The real trick is to identify the root cause of the spike and take corrective action. In my experience, using systemd and ps can be a powerful way to manage CPU usage spikes.

Understanding CPU Usage

To effectively manage CPU usage, you need to understand how Linux reports CPU usage. The top command is a good starting point, but for more detailed analysis, I prefer to use the ps command with various options. For example, to view the top 10 processes consuming the most CPU, I use the following command:

ps -eo pcpu,pid,user,args --sort=-pcpu | head -11

This command displays the percentage of CPU used by each process, along with the process ID, user, and command arguments. Don’t bother with the htop command unless you need a more visual representation of CPU usage - ps is usually enough for me.

Using systemd to Manage Services

systemd is a powerful tool for managing services on Linux systems. By default, systemd starts services in parallel, which can lead to CPU usage spikes during boot time. To mitigate this, I’ve configured systemd to start services sequentially using the systemd-analyze command. For example, to analyze the boot process and identify services that can be optimized, I use the following command:

systemd-analyze blame

This command displays a list of services that are started during boot, along with their respective start times. By optimizing the start order and dependencies of these services, you can reduce CPU usage spikes during boot time. I usually start with the services that take the longest to start and work my way down.

Configuring systemd Services

To further manage CPU usage, you can configure some services to run with limited CPU resources using systemd’s CPUQuota and CPUShares options. For example, to limit the CPU quota of a service to 50%, you add the following lines to the service file:

[Service]
CPUQuota=50%

This ensures that the service does not consume more than 50% of the available CPU resources. In practice, this can be a good way to prevent a single service from hogging all the CPU resources.

Monitoring CPU Usage with ps and systemd

To monitor CPU usage in real-time, I use a combination of ps and systemd. I’ve created a simple script that uses ps to monitor CPU usage and systemd to restart services that exceed a certain CPU threshold. For example:

#!/bin/bash

while true
do
  CPU_USAGE=$(ps -eo pcpu,pid,user,args --sort=-pcpu | head -2 | awk '{print $1}')
  if [ $CPU_USAGE -gt 80 ]; then
    systemctl restart myservice
  fi
  sleep 1
done

This script monitors the top process consuming CPU resources and restarts the myservice service if the CPU usage exceeds 80%. This is where people usually get burned - they don’t monitor their CPU usage, and their services start to fail.

Trade-Offs and Considerations

While using systemd and ps to manage CPU usage spikes is effective, there are some trade-offs to consider. For example, limiting CPU resources for services may impact their performance. Additionally, restarting services that exceed a certain CPU threshold may cause downtime and affect user experience. Therefore, it’s essential to carefully evaluate the CPU usage patterns of your services and configure systemd and ps accordingly. I usually start with a conservative approach and adjust as needed.

For more information on systemd and its configuration options, I recommend visiting the systemd.io website. Additionally, the freedesktop.org website provides a wealth of information on Linux system management and configuration.

Troubleshooting and Debugging

When troubleshooting CPU usage spikes, it’s essential to use the right tools and techniques. The perf command is a powerful tool for analyzing CPU usage and identifying performance bottlenecks. For example, to analyze the CPU usage of a specific process, I use the following command:

perf top -p <pid>

This command displays a list of functions and system calls that are consuming CPU resources, along with their respective execution times. This is usually the first step in identifying the root cause of a CPU usage spike.

linux  systemd  cpu_usage  self-hosting 

See also