ChaosBlade vs chaosd

ChaosBlade is a free, open-source chaos engineering project comprising the ChaosBlade experiment tool and the ChaosBlade-Box platform. Its stated aim is to help enterprises address availability problems during cloud-native adoption. Experiments cover CPU, memory, network, disk, processes, kernels, and files, as well as multilingual applications and Kubernetes resources. They can run through the command-line interface, Kubernetes YAML and CRDs, HTTP server calls, or the Box interface. ChaosBlade-Box provides a shared experiment interface, automates deployment of experiment tools, and can host ChaosBlade and LitmusChaos. It uses Helm for deployment management and integrates Prometheus monitoring. Safety policies support manual stopping or automatic stopping after a configured timeout. Users can download and unpack the CLI on a host, or install the tool and platform on Kubernetes with Helm. The host installation guide lists Java and MySQL requirements; automatic probe installation also needs Ansible and expect. The project supports Kubernetes and is identified as a Cloud Native Computing Foundation sandbox project.

Chaosd is free Apache-2.0 software for injecting and recovering faults on physical machines. It can create process, network, CPU or memory pressure, disk, and host faults. Use it as a command-line tool or run it as a background service controlled through HTTP requests. Chaosd is deployed separately from Chaos Mesh; PhysicalMachineChaos experiments require service mode on each targeted physical or virtual machine. The documented environment requires glibc 2.17 or later, and the binary download described is a Linux amd64 tarball. Network experiments using tc or iptables require sudo or root privileges. Deployment prerequisites include tc, ipset, and iptables; source builds additionally list stress-ng and byteman. Experiments can be searched and recovered by UID. TLS certificates are shown for service security, and running without generated certificates is not recommended.