About the SDN-C Clustering

The following diagram illustrates the desired SDN-C cluster deployment (as described in SDNC-163: Deploy a SDN-C high availability environment - Kubernetes ClosedPreview) from POD and SERVICE view, which will have:

• 2 DB pods

• 3 sdnc (ODL) pods

• 1 admin (portal) pod

• 1 dgbuilder pod

SDN-C DB Clustering details

See details from SDN-C DB (MySQL) Clustered Deployment.

Deployment Setup Steps

You can use Rancher or kubeadm to deploy your Kubernetes cluster:

• Deployment with Kubernetes Cluster Configured by Rancher or

• Deploying Kubernetes Cluster with kubeadm

Additional Information

SDN-C Startup Order

Troubleshoot SDN-C pods are not start up in order

Usually, this is caused by init-container malfuntion. You can troubleshoot it by the following steps:

• Do a describe of the pod using command kubectl describe pod <pod-name>, to ensure the Init Container section is presented as your defined in your yaml template.

• If it does not exist, it is possible that your Kubernetes version supports a different format for the init-containers:

Restart Dead Instance (start a new one in its place)

When a pod is dead, Kubernetes automatically starts a new one to replace the dead pod.

Examples:

• Pod-level restart:

Example of restarting dead instance by deleting pod

sdnc-0 pod runs fine at the beginning

Open

manually delete the sdnc-0 pod

Open

once the sdnc-0 pod is fully terminated, a new sdnc-0 pod will be initialized

Open

The new sdnc-0 pod will be brought to running status to replace the initial sdnc-0

Open

The new and initial sdnc-0 pod has the same name (although with different internal IP)

• Container-level restart:

Example of restarting dead instance by stopping karaf inside SDNC pod

Pod sdnc-0 is in Running STATUS with 0 RESTARTS

Open

Enter pod container and stop Karaf

Open

Kubernetes discovers pod process stopped and restarts the container

Open

sdnc-0 pod will be brought up with the restart container and running with 2 out of 2 container READY

Open

Description of the pod

ubuntu@sdnc-k8s-2:~$ kubectl describe pod sdnc-0 -n onap-sdnc
Name: sdnc-0
Namespace: onap-sdnc
Node: sdnc-k8s-2/10.147.90.6
Start Time: Fri, 24 Nov 2017 16:41:42 +0000
Labels: app=sdnc

...

Containers:
sdnc-controller-container:
Container ID: docker://d3cd42925952d43fa5d6cca359c52d7033fa8bdf5e31881eea424664dde0f634
Image: nexus3.onap.org:10001/onap/sdnc-image:1.2-STAGING-latest
Image ID: docker-pullable://nexus3.onap.org:10001/onap/sdnc-image@sha256:87171a78f8f3090a5f31b63bfead858ec80d220f3ddb4a08e5c98d02cc5edeca
Ports: 8181/TCP, 8101/TCP
Command:
/opt/onap/sdnc/bin/startODL.sh
State: Running
Started: Fri, 24 Nov 2017 17:37:41 +0000
Last State: Terminated
Reason: Completed
Exit Code: 0
Started: Fri, 24 Nov 2017 16:42:15 +0000
Finished: Fri, 24 Nov 2017 17:37:37 +0000
Ready: True
Restart Count: 1

...

Source Code

The source code is currently temporary shared through gerrit topic SDNC-163 until it is commited into gerrit.