In my case, I needed to recreate a group of many pods, controlled by different deployments/statefulsets with restartPolicy: Always

1. Assign label for pods. On deployment/statefulset/daemonset manifest add label at the .spec.template.metadata.labels.
2. Apply/recreate resources (in any way convenient for you). Then check if the changes have been applied:

# kubectl get po -l "app_group=my_grp"

3. Recreate pods by deleting them (according to their controllers restart policy .spec.template.spec.restartPolicy):

# kubectl delete po -l "app_group=my_grp"

Make sure that the old pods are terminating and new ones should be created instead:

# kubectl get po -w

But there's probably a better way.