When abnormal reboot or crash happened to openshift-node or docker. All of the pods will be recreated after openshift backing online. And after OCP3.10, ovs and sdn had been placed into pod instead of local services.

Do we have any mechanism to let ovs/sdn pods always stared before app pods ? If the app pods started at first, it can not assign network settings due to openvswitch is not running. It will get below error until we recreate it after ovs/sdn pods is running.

Warning  FailedCreatePodSandBox  4m                 kubelet, 311master.example.com        Failed create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" network for pod "xxxxxxxxxxxxxxxxxx": NetworkPlugin cni failed to set up pod xxxxxxxxxxxxxxxxxx" network: OpenShift SDN network process is not (yet?) available
Warning  FailedCreatePodSandBox  4m                 kubelet, 311master.example.com        Failed create pod sandbox: rpc error: code = Unknown desc = failed to set up sandbox container "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" network for pod "webconsole-6889b6c96b-m82nw": NetworkPlugin cni failed to set up pod "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" network: OpenShift SDN network process is not (yet?) available
Normal   SandboxChanged          4m (x3 over 4m)    kubelet, 311master.example.com        Pod sandbox changed, it will be killed and re-created.
Warning  NetworkFailed           4m                 openshift-sdn, xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx  The pod's network interface has been lost and the pod will be stopped. 


The customer and I think that this should be avoided in any of situations and environments.

Looking forward to your reply, thank you.