* Fix random failure in debug: var=result.content|from_json

* netchecker agents are deployed on all k8s-cluster group members

* reducing limits/requests is not enough, switching to n1-standard-2

* gce_centos7 need more cpu