endpoints ip address disappeared and re-come back around every 20-30 minutes

[sonyafenge]

What happened:
run kube-up.sh and start-kubemark.sh to start kubemark cluster with 2 apiserver and found endpoint ip address display correctly after kubemark started, but disappeared some time later and keeping this disappear/comeback cycle.
What you expected to happen:
endpoint ip address always display correctly
How to reproduce it (as minimally and precisely as possible):

1. run the command line below to start kube-up and kubemark

export MASTER_DISK_SIZE=300GB MASTER_ROOT_DISK_SIZE=100GB KUBE_GCE_ZONE=us-west2-b MASTER_SIZE=n1-highmem-32 NODE_SIZE=n1-highmem-8 NUM_NODES=2 NODE_DISK_SIZE=200GB GOPATH=$HOME/go KUBE_GCE_ENABLE_IP_ALIASES=true KUBE_GCE_PRIVATE_CLUSTER=true CREATE_CUSTOM_NETWORK=true KUBE_GCE_INSTANCE_PREFIX=demoetcd KUBE_GCE_NETWORK=demoetcd APISERVERS_EXTRA_NUM=0 WORKLOADCONTROLLER_EXTRA_NUM=0
./cluster/kube-up.sh
export MASTER_DISK_SIZE=300GB MASTER_ROOT_DISK_SIZE=100GB KUBE_GCE_ZONE=us-west2-b MASTER_SIZE=n1-highmem-32 NODE_SIZE=n1-highmem-8 NUM_NODES=2 NODE_DISK_SIZE=200GB GOPATH=$HOME/go KUBE_GCE_ENABLE_IP_ALIASES=true KUBE_GCE_PRIVATE_CLUSTER=true CREATE_CUSTOM_NETWORK=true KUBE_GCE_INSTANCE_PREFIX=demoetcd KUBE_GCE_NETWORK=demoetcd APISERVERS_EXTRA_NUM=1 WORKLOADCONTROLLER_EXTRA_NUM=0
./test/kubemark/start-kubemark.sh

2. check endpoints status and get correct endpoints ip address

$ kubectl --kubeconfig=/home/sonyafenge/go/src/k8s.io/arktos-perftest/test/kubemark/resources/kubeconfig.kubemark get endpoints -A
NAMESPACE     NAME                   ENDPOINTS                      AGE    SERVICEGROUPID
default       kubernetes             10.40.0.5:443,10.40.0.12:443   102m   0,1
kube-system   kube-scheduler         <none>                         102m   <none>
kube-system   kubernetes-dashboard   10.78.91.64:8443               102m

3. apply apiserver datapartion and wait around 20 minutes to check endpoints again, then get :

$ kubectl --kubeconfig=/home/sonyafenge/go/src/k8s.io/arktos-perftest/test/kubemark/resources/kubeconfig.kubemark get endpoints -A
NAMESPACE     NAME                   ENDPOINTS                      AGE   SERVICEGROUPID
default       kubernetes             10.40.0.5:443,10.40.0.12:443   89m   0,1
kube-system   kube-scheduler         <none>                         89m   <none>
kube-system   kubernetes-dashboard                                  89m

Anything else we need to know?:

Environment:

• Arktos version (use kubectl version):

commit b82af68a72d74bd86f652cf5bb7318beee547601 (upstream/master)

• Cloud provider or hardware configuration:
• OS (e.g: cat /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Network plugin and version (if this is a network-related bug):
• Others: