Requirements
This topic describes the resource permissions, fault compatibility matrix, and the permissions required to execute Linux chaos experiments.
Resource consumption
The infrastructure consumes minimal system resources in an idle state, when no experiment is being executed. For example, in a GCP e2-micro VM instance with 2 vCPU and 1 GB of memory that runs Ubuntu 22.04 operating system, the average resource consumption was found to be as follows:
- CPU usage: 0.05%
- Memory usage: 1.5%
- Disk storage consumption: 25 MB
- Bandwidth consumption: 0.15 KB/s
Fault compatibility matrix
The faults have been tested for compatibility in the following Linux OS distributions:
Stress faults (cpu, memory, disk IO) | Network faults (loss, latency, corruption, duplication) | DNS faults (error, spoof) | Process faults (process kill, service restart) | Time chaos | Disk fill | |
---|---|---|---|---|---|---|
Ubuntu 16+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Debian 10+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
CentOS 7+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
RHEL 7+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
Fedora 30+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
openSUSE LEAP 15.4+ / SUSE Linux Enterprise 15+ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
On-premise VMs (VMware VMs)
Linux OS
Chaos agent deployment model | Native Chaos Agent on Each VM (system service within Target Linux Machine) | Centralized Chaos agent on Kubernetes (leverage VMware Tools to inject chaos processes inside guest VM ) |
---|---|---|
Connectivity requirements from agent |
|
|
Connectivity requirements from VM/cluster/app |
|
|
Access requirements for agent install |
|
|
Access requirements for basic chaos experiments |
|
|
Access requirements for advanced chaos experiments |
|
|
Supported chaos faults |