Preparing the deployment of Unified Agent

Before you deploy the Unified Agent in Kubernetes environments, make sure that your environment meets the prerequisites.

General prerequisites

Docker V18 or higher version
Kubectl client on the environment from where you are installing
If your system runs on SELinux, disable SElinux or set it to permissive mode when installing the Unified Agent.

System requirements

For detailed system requirements of Unified Agent and plug-ins, see System requirements for agents, data collectors, and plug-ins.

Special prerequisites for plug-ins

Some plug-ins have special requirements. Before you install the Unified Agent and enable these plug-ins, you must make sure to complete the following tasks.

IBM API Connect plug-in

The API Connect toolkit is required to provide CLI commands to register the plug-in.
The Unified Agent must be deployed in the same Kubernetes environment as IBM API Connect.
This plug-in leverages Kubernetes Metrics API to get the pod CPU and Memory usage, so metrics server is required to be deployed in the cluster.

NGINX plug-in

Make sure that the NGINX monitoring interface is enabled.

Run the following command on the machine where you want to deploy the Unified Agent and check whether the NGINX monitoring interface is enabled:
```
http://pod_ip_or_nginx_host:18080/nginx_status
```
where pod_ip_or_nginx_host is the fully qualified hostname of the NGINX server.
If no status is returned, the NGINX monitoring interface is not enabled.
The NGINX monitoring interface requires loading the ngx_http_stub_status_module module. This module helps in collecting basic performance metrics. You need only to get the Kubernetes SERVICE or POD IP and verify that the management interface is enabled. For example, http://NGINX_service_or_node_or_pod_ip:18080/nginx_status. See the following command examples to help you determine whether NGINX workloads are running and confirm that their management interfaces are enabled.
```
 kubectl get po -n kube-system -o wide |grep nginx
```
The command returns output similar to the following example:
```
 nginx-ingress-controller-jx8vb      1/1     Running   0     5h   
 10.1.253.201   9.42.75.39
```

Verify the NGINX management interface status:

curl http://10.1.253.201:18080/nginx_status

You can see the following example output:

Active connections: 9
server accepts handled requests
 5372 5372 22532
Reading: 0 Writing: 2 Waiting: 7

Because the pod IP address can change, you can optionally create a service that points to the POD to get a static IP address. The following commands help create a service configuration:

kubectl describe po nginx-ingress-controller-jx8vb -n kube-system

Name:              nginx-ingress-controller-jx8vb 
Namespace:         kube-system 
Priority:          0 
PriorityClassName: <none> 
Node:              9.42.75.39/9.42.75.39 
Start Time:        Fri, 28 Sep 2018 23:46:39 -0400 
Labels:            app=nginx-ingress-controller

Note: Here the selector is app=nginx-ingress-controller. It might differ in your Kubernetes environment.

Create a service resource file (nginx-status.yaml) as shown here:

{
  "apiVersion": "v1",
   "kind": "Service",
   "metadata": {
     "name": "nginx-status",
     "namespace": "kube-system",
     "labels": {
       "app": "nginx-status"
     }
   },
   "spec": {
     "ports": [
       {
         "name": "nginx-status",
         "protocol": "TCP",
         "port": 18080,
         "targetPort": 18080
       }
     ],
     "selector": {
       "app": "nginx-ingress-controller"
     },
     "type": "ClusterIP",
     "sessionAffinity": "None"
   }
}

Note: Open port 18080 for NGINX status access.

Create the Kubernetes service resource by using the service resource file that you create, and obtain the service ip.

 > kubectl create -f nginx-status.yaml
 > kubectl describe svc nginx-status -n kube-system

 Name:              nginx-status
 Namespace:         kube-system
 Labels:            app=nginx-status
 Annotations:       <none>
 Selector:          name=nginx-ingress-controller
 Type:              ClusterIP
 IP:                10.0.0.243
 Port:              nginx-status  18080/TCP
 TargetPort:        18080/TCP
 Endpoints:         9.37.22.210:18080
 Session Affinity:  None
 Events:            <none>

Run curl http://10.0.0.243:18080/nginx_status. You can give the NGINX management interface URL http://10.0.0.243:18080/nginx_status to the Helm Chart configuration. In on-premises installations, this module is not enabled by default. It must first be built and then enabled with the configuration parameter --with-http_stub_status_module. See the NGINX documentation for enablement.