Hands-On Microservices with Kubernetes

Node failure

When a node fails, all the pods on the node will become unavailable and Kubernetes will schedule them to run on other nodes in the cluster. Assuming you design your system with redundancy in place and the failed node is not a SPOF, the system should recover automatically. If the cluster has just a few nodes, then the loss of a node can be significant to the cluster's ability to handle traffic.