Installing services with Cloud Pak for Data 3.5

Compared to a standalone Cloud Pak for Data product, Cloud Pak for Data System comes preconfigured with OpenShift cluster, Portworx, and Cloud Pak for Data base software. Starting with the System version 1.0.7.5, the services (add-ons) are installed by using the cpd-cli command. Use this procedure if Cloud Pak for Data installed on the system is of version 3.5 (Cloud Pak for Data System versions 1.0.7.5 and later).

Before you begin

Some services that run on Cloud Pak for Data require specific settings on the nodes in the cluster. To ensure that the cluster has the required settings for these services, an operating system administrator with root privileges must review and adjust the settings on the appropriate nodes in the cluster, as described in https://www.ibm.com/support/knowledgecenter/en/SSQNUZ_3.5.0/cpd/install/node-settings.html.

About this task

In Cloud Pak for Data version 3.5, only the download-images and push-images commands are common to all services. The remaining part of the installation procedure is service-specific, and you need to consult the documentation for the specific service that you are installing. For the Cloud Pak for Data stand-alone product, it is recommended to run the installation from and external workstation. In the Cloud Pak for Data System environment, you can use e1n1-1-control instead. The cpd-cli tool is pre-installed there and ready to use.

Following is the overview the service installation process in an air-gapped environment as documented for stand-alone Cloud Pak for Data:

Begin Alt text. Install process on an air-gapped cluster. This process is explained in the preceding text. End Alt text. — Figure 1. Flow when running the `cpd-cli` command on an air-gapped cluster

Cloud Pak for Data defines the modular installation components as follows:

Assembly

A particular Cloud Pak for Data service (and its dependent assemblies) to deploy to a project in Red Hat OpenShift. Assemblies for Cloud Pak for Data control plane and services are stored on a public IBM file server, and do not have to be installed at the same time.

Repository configuration

A server definition YAML file that you download to your Linux or Mac OS client workstation and then customize. It specifies:

URLs and credentials for the file server to download Helm charts from.
URLs and credentials for the registry server to download images from.

The default server definition file is repo.yaml.

When you run the cpd-cli download command from a client workstation, it downloads all of the images and Helm charts from a public IBM file server to the client workstation. Next, you move the Cloud Pak for Data command-line interface and the downloaded files to the Cloud Pak for Data System node e1n1. From there, run the cpd-cli command to push the images to your registry server. Lastly, you run the cpd-cli installation command, which deploys the Helm charts to a project in your OpenShift cluster.

In the Cloud Pak for Data System environment, the services images should be pushed to the following registry docker-registry.default.svc:5000/zen.

In the following procedure, you download the images to your local workstation (your laptop or a local machine), from which you can connect directly to the Cloud Pak for Data System cluster.

Procedure

On a local workstation that has internet access:
1. Download the cpd tool and the oc tool installer from IBM Passport Advantage. For more details, see https://www.ibm.com/support/knowledgecenter/SSQNUZ_3.5.0/cpd/install/installation-files.html.
2. Get the entitlement key.
  You can get your entitlement license API key from the Container software library on My IBM.
  After you order a service, an entitlement key for the software is associated with your My IBM account. To get the entitlement key:
  1. Log in to Container software library on My IBM with the IBM ID and password that are associated with the entitled software.
  2. On the Get entitlement key tab, select Copy key to copy the entitlement key to the clipboard.
  3. Save the API key in a text file.
3. Edit the server definition file repo.yaml that is in the downloaded package. This file specifies the repositories for the cpd command to download the installation files from. Make the following changes to the file:
  
  Parameter Value
  
  username Specify cp.
  
  apikey Specify your entitlement license API key.
4. Change to the directory where you extracted the Cloud Pak for Data installation command-line interface.
5. Run the following command to download the required files to your local machine; ensure that the download path directory name is same as assembly name:
```
./cpd-cli preload-images --assembly <assembly name> --action download -r repo.yaml --download-path=<dir>
```
  Example:
```
./cpd-cli preload-images --assembly wsl --action download -r ../repo.yaml --download-path=wsl
```
On control node e1n1, create an /opt/ibm/appliance/storage/platform/appmgt/installers/services directory for the assemblies:
```
[root@e1n1 ~]# mkdir /opt/ibm/appliance/storage/platform/appmgt/installers/services
```
Transfer the files to the control node e1n1 in the Cloud Pak for Data System cluster under /opt/ibm/appliance/storage/platform/appmgt/installers/services. Ensure that the folder structure is preserved if the files are not in the tar.gz format.

Parameter	Value
`username`	Specify `cp`.
`apikey`	Specify your entitlement license API key.

Untar the downloaded files if needed:

[root@e1n1 ~]# tar -xzvf <the-tar-gz-file>

Copy cpd-cli binary to the same services directory:

[root@e1n1 ~]# cp /opt/ibm/appliance/storage/platform/appmgt/installers/cpdimages/cpd-cli  /opt/ibm/appliance/storage/platform/appmgt/installers/services

ssh into e1n1-1-control:
```
[root@e1n1 ~]#  ssh  e1n1-1-control
```

Push the images to registry:

./cpd-cli preload-images   
--assembly Assembly_name 
--arch Cluster_architecture 
--action push 
--ask-push-registry-credentials 
--load-from Image_directory_location 
--transfer-image-to Registry_location

Example:

[root@e1n1-1-control ~]# ./cpd-cli preload-images --action push --load-from ./wsl --assembly wsl --transfer-image-to=docker-registry.default.svc:5000/zen

Consult the Cloud Pak for Data documentation pertaining to the service that you want to install and follow the procedure for installing on air-gapped clusters.

Note: When installing services, it is recommended to use portworx-shared-gp2 or portworx-shared-gp3 storage class for high-availability, and to avoid data loss with single replica storage volume. Do not use portworx-shared-gp.