Write script to automate the Gather info and files step

[bmpotter]

• [x] Automate as many of the steps in Gather the necessary information and files as possible.
• [x] We also might want to want to add the step of pulling down agent-install.sh from the anax repo and make it executable.
• [x] Also, provide the option to tar up the files, so they can more easily be transferred to edge nodes. This is also needed for the batch node install process.

Not sure yet where this script should live.

[t-fine]

For now I added the script to the tools directory in the open-horizon git repo, but that was just so I could save it in GitHub.

https://github.com/t-fine/examples/blob/gather-script/tools/edgeDeviceFiles.sh

• [x] add flag to specify distribution (xenial or bionic) if 64-bit-ARM or x86_64-LINUX are specified

[t-fine]

Requirements:

• cloudctl and kubectl are installed (checked in first function; will fail if not installed)
• ICP_URL, USER, and PW must be set for cloudctl login ... command (also checked in first function)

[t-fine]

--------------- Knowledge Center Doc Changes -------------------

https://www-03preprod.ibm.com/support/knowledgecenter/SSFKVV_3.2.1/devices/installing/install.html

Gather the necessary information and files

The script in this section only needs to be executed once for each edge node device type, but must be performed on an x86 host (Linux, macOS, or Windows). If your edge device is a different architecture (for example arm or arm64), then run the script in this section on an x86 host and copy the resulting tar file to your edge device.

1. Install the IBM Cloud Pak CLI (cloudctl) and the Kubernetes CLI (kubectl) by following the steps in Installing cloudctl (LINK: https://www.ibm.com/support/knowledgecenter/SSBS6K_3.2.1/manage_cluster/install_cli.html ).

Alternatively, you can install kubectl using your package manager. For more information about installing kubectl with your package manager, see Install and set up kubectl (LINK: https://kubernetes.io/docs/tasks/tools/install-kubectl/ ).

2. After cloudctl is installed, ask your IBM cluster administrator for the URL for the cluster, username, and password. Export these values as environment variables:

   export CLUSTER_URL=<cluster-url>
   export USER=<username>
   export PW=<password>

3. Download the latest version of the edgeDeviceFiles.sh:

   curl -O https://raw.githubusercontent.com/open-horizon/examples/master/tools/edgeDeviceFiles.sh && \
   chmod +x edgeDeviceFiles.sh

4. Run the script to gather the necessary files to register an edge node with the edgeDeviceFiles.sh script.

   • Supported <edge-device-type> values: 32-bit-ARM , 64-bit-ARM , x86_64-Linux , macOS
   • -t: create an agentInstallFiles-<edge-device-type>.tar.gz file containing all gathered files. If this flag is not set, the gathered files will be placed in the current directory.
   • -k: create a new $USER-API-Key. If this flag is not set, the existing api keys will be checked for $USER-API-Key and creation will be skipped if one exists.
   • -d <distribution>: when installing on 64-bit-ARM or x86_64-Linux: you can specify -d xenial for the older version of Ubuntu, instead of the default bionic. The -d flag is ignored with all other values.

./edgeDeviceFiles.sh <edge-device-type> -t 

[Rene-Ch1]

This might be related to https://github.ibm.com/Edge-Fabric/docs/issues/228

[bmpotter]

Tested/used this multiple times and it worked like a champ!