Pod restart behaviour

[merorafael]

Hi folks, I noticed a strange behaviour during a kubectl-argo-rollouts restart (doc) command execution. Considering this doc, Argo Rollouts uses maxUnavailable to determine how many pods will be terminated by time. Considering the value 0% for maxUnavailable parameter, the pods are restarted once per time (following the documentation). However, the controller is not waiting for the pod to be ready before following with the other changes, resulting in downtime in the application. Is there a way to avoid this behaviour, and the Argo Rollouts controller waits for the ready pod's status before terminating another pod?

[kostis-codefresh]

Do you have liveness and readiness probes in your pods?