Currently, all k8s master components (kube-scheduler, kube-controller-manager, kube-apiserver) only log to their respective container's stdout. These logs are destroyed when the pod is recreated.
Instead, these components should be configured to log to disk directly using the --log-dir and --logtostderr=false options.
Currently, all k8s master components (kube-scheduler, kube-controller-manager, kube-apiserver) only log to their respective container's stdout. These logs are destroyed when the pod is recreated.
Instead, these components should be configured to log to disk directly using the
--log-dirand--logtostderr=falseoptions.