Pod kill
Proves readiness probes, disruption budgets and replicas do their job.
⚠ Guide
Kubernetes chaos testing, one Helm install away
Kubernetes chaos testing means killing pods, draining nodes, starving resources and slowing network traffic to check that self-healing actually works. Start with pod kills against one deployment, confirm disruption budgets and probes, then widen the blast radius.
Experiments worth running
Node drain and loss
Proves scheduling, anti-affinity and cluster autoscaling.
Network latency
Adds delay between services to find missing timeouts.
CPU and memory stress
Proves limits, requests and horizontal pod autoscaling.
Install
helm install omggaas omggaas/agent --set token=$OMG_TOKEN
Safe by design
Every fault Oh My GDaaS runs is aimed at infrastructure you have proved you own, through a DNS record, a scoped cloud role or a deployed agent. Every run has a blast-radius limit, a hard time limit and an abort button. We are not a booter, stresser or DDoS service and never will be.