Deploying API Manager with Kubernetes or OpenShift Resources

Follow the instructions below to use Kubernetes (K8s) or OpenShift resources for container-based deployments of WSO2 API Manager (API-M).

In the context of this document, < KUBERNETES_HOME> refers to a local copy of the wso2/kubernetes-apim Git repository that includes common resources that can be used for Kubernetes and OpenShift.

Checkout the WSO2 kubernetes-apim repository using git clone:

git clone https://github.com/wso2/kubernetes-apim.git
git checkout tags/v2.1.0-2

Pull required Docker images from the WSO2 Docker Registry using docker pull:
```
docker login docker.wso2.com

docker pull docker.wso2.com/wso2am-analytics-kubernetes:2.1.0
docker pull docker.wso2.com/wso2am-kubernetes:2.1.0
docker pull docker.wso2.com/apim-rdbms-kubernetes:2.1.0
```
You can also build the Docker images by following the guide in the <KUBERNETES_HOME>/base/README.md. In addition, note that the same images can be used for OpenShift.
Copy the Images into Kubernetes/OpenShift nodes or to a Registry.
Copy the required Docker images over to the Kubernetes Nodes.
For example:
1. Use docker save to create a TAR file of the required image.
2. scp the TAR file to each node.
3. Use docker load to load the image from the copied TAR file on the nodes.
  Alternatively, if a private Docker registry is used, transfer the images there.
Deploy Kubernetes/OpenShift Resources.
Before you begin the deployment, make sure that you have the following prerequisites.
- Set up Network File System (NFS) to deploy any pattern.
  NFS is used as the persistent volume for API Manager servers. As a result, setting up NFS is required to deploy any pattern. Therefore, you need to complete the following:
  1. Update the NFS server IP in <KUBERNETES_HOME>/pattern-X/artifacts/volumes/persistent-volumes.yaml
  2. Create the required directories in the NFS server for each pattern as mentioned in the <KUBERNETES_HOME>/pattern-X/artifacts/volumes/persistent-volumes.yaml
    For example, for pattern-1, create the directories as /exports/pattern-1/apim
- It is recommended to use a MySQL or any database cluster in a production environment.
  Only one MySQL container is used with host path mount in these deployments.
Deploy Kubernetes/OpenShift Resources:

The following instructions have been tested on OpenShift v3.6.0 and Kubernetes v1.6.1 and NFS is tested in Kubernetes v1.6.1.

Deploy a pattern on Kubernetes
1. Create a namespace named wso2.
```
kubectl create namespace wso2
```
2. Create a service account named wso2svcacct in the wso2 namespace.
```
kubectl create serviceaccount wso2svcacct -n wso2
```
3. Deploy any pattern by running the deploy-kubernetes.sh script that is inside the pattern folder (<KUBERNETES_HOME>/pattern-X/artifacts directory).
```
./deploy-kubernetes.sh
```
4. Access the management console using the following command to list ingresses in the deployment.
```
kubectl get ingress
```
  Add relevant hosts and IP addresses to the /etc/hosts file.
  The following are sample access URLs. However, note that this will vary based on the pattern that you are using.
If required, undeploy a pattern on Kubernetes
You can undeploy any pattern by running the undeploy-kubernetes.sh script that is inside the pattern folder (<KUBERNETES_HOME>/pattern-X/ directory).
```
./undeploy-kubernetes.sh
```
Deploy a pattern on OpenShift
- Step 1 - Configure OpenShift
- Step 2 - Deploy the pattern
Step 1 - Configure OpenShift
1. Create a user named admin and assign the user to the cluster-admin role.
  This user with the cluster-admin role is used to deploy the OpenShift artifacts.
```
oc login -u system:admin
oc create user admin --full-name=admin
oc adm policy add-cluster-role-to-user cluster-admin admin
```
2. Create a new project named wso2.
```
oc new-project wso2 --description="WSO2 API Manager 2.1.0" --display-name="wso2"
```
3. Create a service account named wso2svcacct in the wso2 project.
```
oc create serviceaccount wso2svcacct
```
4. Assign the service account named wso2svcacct to the anyuid security context constraint
```
oc adm policy add-scc-to-user anyuid -z wso2svcacct -n wso2
```
5. Define the service account in the deployment artifact YAML file.
  For example, see the wso2apim-manager-worker-deployment.yaml file.
```
serviceAccountName: "wso2svcacct"
```
Step 2 - Deploy the pattern
1. Deploy any pattern by running the deploy-openshift.sh script inside the pattern folder (<KUBERNETES_HOME>/pattern-X/ directory).
```
./deploy-openshift.sh
```
2. Access the Management Console using the following command to list the routes in the deployment.
```
oc get routes
```
  Add relevant hosts and IP addresses to the /etc/hosts file.
  The following are sample access URLs. Note that this varies based on the pattern that you are using.
If required, undeploy a pattern on OpenShift
You can undeploy any pattern by running the undeploy-openshift.sh script that is inside the pattern folder (<KUBERNETES_HOME>/pattern-X/directory).
```
./undeploy-openshift.sh
```
Customize the deployment (If required).

Configurations are bound with the wso2 namespace. Therefore, if you are changing the hostnames or the namespace, do the following:
1. Change wso2.svc to <namespace>.svc in all the configuration files.
2. Update the KUBERNETES_NAMESPACE parameter with the correct namespace in all the <API-M_HOME>/ repository/conf/axis2/ axis2.xml files.
3. Update Docker base images.
  Use a CA signed certificate and update the client-truststore.jks and wso2carbon.jks files which are in the following location.
  https://github.com/wso2/kubernetes-apim/tree/2.1.0-nfs/base/apim/files.

Deploying API Manager with Kubernetes or OpenShift Resources

Deploy a pattern on Kubernetes

Deploy a pattern on OpenShift

Step 1 - Configure OpenShift

Step 2 - Deploy the pattern