Description
The problem runs on social-network. Simulate a fault where nodes are tainted but pods lack corresponding tolerations in the social network application. Worker nodes are tainted with sre-fault=blocked:NoSchedule, but the deployment `user-service` has a toleration for a different key (dummy-key), causing pods to be unschedulable and remain in Pending state. Inject the fault by adding taints to nodes without configuring corresponding tolerations in pod specifications.
Tags
Cloud Management System Failure, social-network