A cluster needs to maintain three interchangeable web replicas after a node fails. Which workload object is appropriate?

Published
  1. PersistentVolume
  2. A standalone Pod only
  3. ConfigMap
  4. Deployment

Comments — 0