see
https://lists.onap.org/g/onap-discuss/topic/onap_oom_portal/23245878?p=,,,20,0,0,0::recentpostdate%2Fsticky,,,20,2,0,23245878

Udhay,
Hi, Your issue is most likely Load balancer (service) related.
Kubadm does not ship with a default --type=LoadBalancer
The rancher RI for ONAP Kubernetes deployment ships with a built in load balancer – portal is the only LB service currently in ONAP. Without this load balancer you would need to manually configure your own native load balancer – like a classic ELB in Amazon AWS for example.

Since you are using kubeadm to run your kubernetes cluster instead of the Rancher 1.6.14/1.6.18 RI – I would expect that the lack of a configured LB would not distribute the request to the proper nodeport on a particular VM on your cluster

Everyone,
A good way to test whether any portal issue is LB related - would be to expose the built in Grafana (Thanks Mandeep for pointing this out) – monitoring service as a LoadBalancer
https://wiki.onap.org/display/DW/Cloud+Native+Deployment#CloudNativeDeployment-GrafanaDashboards
In this case the LB service will navigate to the port (just like portal would) regardless of which node I launch from master0-8

http://master1.onap.info:30928/dashboard/db/cluster?orgId=1

1. expose the nodeport
   kubectl expose -n kube-system deployment monitoring-grafana --type=LoadBalancer --name monitoring-grafana-client
   service "monitoring-grafana-client" exposed
2. get the nodeport pod is running on
   kubectl get services --all-namespaces -o wide | grep graf
   kube-system monitoring-grafana ClusterIP 10.43.44.197 <none> 80/TCP 7d k8s-app=grafana
   kube-system monitoring-grafana-client LoadBalancer 10.43.251.214 18.222.4.161 3000:32628/TCP 15s k8s-app=grafana,task=monitoring
3. get the cluster vm DNS name
   ubuntu@ip-10-0-0-169:~$ kubectl get pods --all-namespaces -o wide | grep graf
   kube-system monitoring-grafana-997796fcf-7kkl4 1/1 Running 0 5d 10.42.84.138 ip-10-0-0-80.us-east-2.compute.internal

From: M, UdhayaChandran <udhayachandran.m@verizon.com>
Sent: Wednesday, July 11, 2018 10:26 AM
To: Michael O'Brien <Frank.Obrien@amdocs.com>
Subject: ONAP_OOM Portal

Hi Micheal,

Good day,

I am trying to bring up Portal alone as a standalone component using OOM with kubeadm.

All the pods are up and running, but i could see the services of portal-app external_ip's to be in the pending state.

when i tried to bring up Portal using Rancher, the external Ip is getting allocated, and works good.

Kindly let me know, do i need to configure any external settings to make the portal-app to assign external-ip's.

regards,
Udhay