Chaos Engineering for GKE Autopilot with Steadybit

Google Kubernetes Engine (GKE) Autopilot is a fully managed Kubernetes platform that simplifies infrastructure management, allowing teams to focus on application deployment. While this locked-down design limits chaos engineering, Steadybit, as an official Google Autopilot partner, offers a solution by providing integration capabilities for chaos engineering practices within these managed environments. By utilizing a central control plane and agents, Steadybit enables container-level fault injections similar to those in standard clusters, although host-level and node-level attacks remain unavailable due to GKE Autopilot's restrictions. The installation process involves configuring an allowlist and deploying the Steadybit agents and extensions using Helm Charts, allowing teams to run various container-level chaos experiments. This partnership allows users to leverage both the ease of GKE Autopilot and enhanced application reliability through Steadybit's tools, with additional support for technologies like Kafka and Splunk.