Simulate Stress Scenarios

This document describes how to use Chaosd to simulate stress scenarios. This feature generates CPU or memory stress on the host using stress-ng. You can create stress experiments either in command-line or service mode.

Before creating stress experiments, you can run the following command to view the stress experiment types supported by Chaosd:

chaosd attack stress --help

Currently, Chaosd supports creating CPU stress experiments and memory stress experiments.

To create experiments using the service mode, you need to run Chaosd in the service mode and then send a POST HTTP request to the /api/attack/stress path of the Chaosd service:

chaosd server --port 31767

curl -X POST 172.16.112.130:31767/api/attack/stress -H "Content-Type:application/json" -d '{fault-configuration}'

For the fault-configuration part in the above command, you need to configure it according to the fault types. For the corresponding parameters and examples, refer to the parameters of each fault type in the following sections.

info

When running an experiment, remember to save the UID information of the experiment. When you want to kill the experiment corresponding to the UID, you need to send a DELETE HTTP request to the /api/attack/{uid} path of Chaosd service.

Simulate CPU stress

Parameters for simulating CPU stress

Configuration item	Abbreviation	Service mode field	Description	Type	Value
action	—	action	Actions of the experiment	string	Set to "cpu"
load	l	load	Specifies the percentage of CPU load per CPU worker. 0 means no CPU utilization, and 100 means full CPU utilization.	int	Range: 0 to 100; Default value: 10.
workers	w	workers	Specifies the number of workers used to create CPU stress.	int	Default value: 1.
options	o	options	The extended parameter of stress-ng, usually not configured.	string	Default value: "".

Simulate CPU stress using the command-line mode

To view the configuration items supported by the CPU stress simulation, run the following command:

chaosd attack stress cpu --help

An example is as follows:

chaosd attack stress cpu --workers 2 --load 10

Simulate CPU stress using the service mode

curl -X POST 172.16.112.130:31767/api/attack/stress -H "Content-Type:application/json" -d '{"load":10, "action":"cpu","workers":1}'

Simulate memory stress

Parameters for simulating memory stress

Configuration item	Abbreviation	Service mode field	Description	Type	Value
action	—	action	Actions of the experiment	string	Set to "mem"
size	s	size	Specifies the size of memory per VM worker.	string	The memory size in B, KB/KiB, MB/MiB, GB/GiB, TB/TiB. If the size is not set, all available memory is used by default.
workers	w	workers	Specifies the number of workers used to create memory stress.	int	Default value: 1.
options	o	options	The extended parameter of stress-ng, usually not configured.	string	Default value: "".

info

The --workers flag for chaosd attack stress mem was restored in chaosd v1.6.0. If you use an earlier version of chaosd, this flag is not available.

Simulate memory stress using the command-line mode

To view the configuration items supported by the memory stress simulation, run the following command:

chaosd attack stress mem --help

An example is as follows:

chaosd attack stress mem --workers 2 --size 100M

When running the experiment, you need to save the uid information of the experiment. When a stress simulation is not needed, you can use recover to terminate the uid-related experiment:

chaosd recover c2bff2f5-3aac-4ace-b7a6-322946ae6f13

Simulate memory stress using the service mode

curl -X POST 172.16.112.130:31767/api/attack/stress -H "Content-Type:application/json" -d '{"action":"mem", "workers":2, "size":"100M"}'