Master stress‑ng: Install, Configure, and Run Comprehensive Linux Stress Tests

What is stress‑ng?

stress‑ng is a powerful Linux stress‑testing utility that can generate load on various system resources such as CPU, memory, disk I/O, network, processes, and file systems. It offers many configurable stressors, making it suitable for performance benchmarking and reliability testing.

Installation

Two installation methods are supported on CentOS 7:

1. yum installation (not the latest version)

yum install -y epel-release.noarch && yum -y update
yum -y install stress stress-ng

2. Build from source (recommended for the newest version)

Download the latest source package, extract it, compile, and install:

wget https://fossies.org/linux/privat/stress-ng-0.16.03.tar.gz
 tar -xzf stress-ng-0.16.03.tar.gz
 cd stress-ng-0.16.03
 make && make install

After compilation, verify the installation: stress-ng -V Typical output:

stress-ng, version 0.16.03 (gcc 4.8.5, x86_64 Linux 3.10.0-957.el7.x86_64)

Command syntax and common options

The basic syntax is: stress-ng [OPTION [ARG]] Key options include: -h, --help: display help --version: show version -t, --timeout seconds : stop after the given time -c, --cpu n : launch n CPU workers (e.g., --cpu 4) -i, --io n : launch n I/O workers -m, --vm n : launch n memory workers -d, --hdd n : launch n workers that perform write/unlink operations --cpu-method method : choose a specific CPU stress method (e.g., matrixprod) --vm-bytes size : amount of memory each VM worker allocates (default 256M) --cpu-load percent : set target CPU load percentage --hdd-bytes size : size of each write operation --class name : run a predefined class of tests

Practical stress‑testing scenarios

1. CPU load test

Run four CPU workers using the matrix multiplication method for 600 seconds:

stress-ng --cpu 4 --cpu-method matrixprod --timeout 600s

Monitor the impact with top before and during the test (screenshots omitted for brevity).

2. Memory load test

Allocate 512 MiB per worker with four workers, keep the allocation for 600 seconds: stress-ng --vm 4 --vm-bytes 512M --timeout 600s Check memory usage with free and top before and during the test.

3. I/O load test

Generate four I/O workers for 600 seconds: stress-ng --io 4 --timeout 600s Observe system load via top or a Prometheus dashboard.

4. Disk write test

Use one write worker that writes 1000 MiB blocks with four I/O workers for 600 seconds:

stress-ng --io 4 --hdd 1 --hdd-bytes 1000M --timeout 600s

5. Combined CPU and I/O test

Run four CPU workers and two I/O workers for 600 seconds:

stress-ng --cpu 4 -i 2 --timeout 600s

6. Adjustable CPU load

Run four CPU workers targeting a 60 % load for 600 seconds:

stress-ng --cpu 4 --cpu-load 60 --timeout 600s

Conclusion

stress‑ng provides a rich set of options for generating realistic load on Linux systems. By installing it (preferably from source for the latest features) and using the examples above, you can quickly create reproducible stress‑testing scenarios for CPU, memory, I/O, and disk subsystems, and verify system behavior with standard monitoring tools.