Install Cloud Pak for Data 3.5 on Red Hat OpenShift

These steps assume that Red Hat OpenShift 4.5 has already been installed on the cluster and a storage class for NFS, OCS or Portworx is available on the cluster.

Prepare installation for Cloud Pak for Data

Log on to the bastion node

Ensure that you're logged on to the bastion node as root.

Use screen to multiplex your terminal

The screen utility allows you to perform a long-running task in a terminal, which continues even if your connection drops. Once started, you can use Ctrl-A D to detach from your screen and screen -r to return to the detached screen. If you accidentally started multiple screens, you can use screen -ls to list the screen sessions and then attach to one of them using screen -r <session_id>. As a best practice, try to have only one screen terminal active to simplify navigation.

screen

Log on to OpenShift

If not already logged on, log on to the bastion node as root; this is the 2nd IP address in the Asset Access Instructions document.

oc login -s https://api.apps.ocp45.uk.ibm.com:6443 -u ocadmin -p passw0rd --insecure-skip-tls-verify=true

Override CRI-O settings

cat << EOF > /tmp/cp4d-crc.yaml
apiVersion: machineconfiguration.openshift.io/v1
kind: ContainerRuntimeConfig
metadata:
  name: cp4d-crio-limits
spec:
  machineConfigPoolSelector:
    matchLabels:
      limits-crio: cp4d-crio-limits
  containerRuntimeConfig:
    pidsLimit: 12288
EOF

oc create -f /tmp/cp4d-crc.yaml

oc label mcp worker limits-crio=cp4d-crio-limits

Override sysctl settings through Tuned

The steps below change the system control parameters required for WKC pods. These changes are typically required, regardless whether the WKC assembly is installed.

cat << EOF > /tmp/42-cp4d.yaml
apiVersion: tuned.openshift.io/v1
kind: Tuned
metadata:
  name: cp4d-wkc-ipc
  namespace: openshift-cluster-node-tuning-operator
spec:
  profile:
  - name: cp4d-wkc-ipc
    data: |
      [main]
      summary=Tune IPC Kernel parameters on OpenShift Worker Nodes running WKC Pods
      [sysctl]
      kernel.shmall = 33554432
      kernel.shmmax = 68719476736
      kernel.shmmni = 16384
      kernel.sem = 250 1024000 100 16384
      kernel.msgmax = 65536
      kernel.msgmnb = 65536
      kernel.msgmni = 32768
      vm.max_map_count = 262144
  recommend:
  - match:
    - label: node-role.kubernetes.io/worker
    priority: 10
    profile: cp4d-wkc-ipc
EOF

oc create -f /tmp/42-cp4d.yaml

Wait until machine configurations have been applied

OpenShift will now apply the machine configurations to all the workers and reboot them one by one. This may take a while to complete and it is best to wait before you continue with the next step.

watch -n 10 'oc get mcp'

Machine configurations have been applied when "worker" line has UPDATED=True and UPDATING=False. You can see that the READYMACHINECOUNT column goes from 0 up to the number of workers. Example:

Every 10.0s: oc get mcp                                                                                                                                                                                                               Sun Sep  6 07:40:44 2020

NAME     CONFIG                                             UPDATED   UPDATING   DEGRADED   MACHINECOUNT   READYMACHINECOUNT   UPDATEDMACHINECOUNT   DEGRADEDMACHINECOUNT   AGE
master   rendered-master-7e735cd7f00167b46e7e663d41634ced   True      False	 False      3              3                   3                     0                      17d
worker   rendered-worker-3bc6374bf0d550a75df4206e1bd94610   False     True	 False      3              1                   1                     0                      17d

If you want to check what is happening with the nodes, you can run the following command:

oc get no

Example:

[root@bastion ~]# oc get no
NAME                        STATUS                     ROLES    AGE   VERSION
master-1.ocp45.uk.ibm.com   Ready                      master   17d   v1.18.3+002a51f
master-2.ocp45.uk.ibm.com   Ready                      master   17d   v1.18.3+002a51f
master-3.ocp45.uk.ibm.com   Ready                      master   17d   v1.18.3+002a51f
worker-1.ocp45.uk.ibm.com   Ready                      worker   17d   v1.18.3+002a51f
worker-2.ocp45.uk.ibm.com   Ready,SchedulingDisabled   worker   17d   v1.18.3+002a51f
worker-3.ocp45.uk.ibm.com   Ready                      worker   17d   v1.18.3+002a51f

Install Cloud Pak for Data (connected)

These steps assume that the OpenShift cluster is connected to the internet. If not, skip to Air-gapped install of Cloud Pak for Data

Download installer

wget https://github.com/IBM/cpd-cli/releases/download/v3.5.2/cpd-cli-linux-EE-3.5.2.tgz -P /tmp/
mkdir -p /nfs/cpd
tar xvf /tmp/cpd-cli-linux-EE-3.5.2.tgz -C /nfs/cpd
rm -f /tmp/cpd-cli-linux-EE-3.5.2.tgz

Obtain your entitlement key for the container registry

Login here: https://myibm.ibm.com/products-services/containerlibrary, using your IBMid. Then copy the entitlement key.

Apply key to the repo.yaml file

Insert the entitlement key after the apikey: parameter in the /nfs/cpd/repo.yaml file. Please make sure you leave a blank after the :.

Install Cloud Pak for Data - individual modules

Use the steps below if you want to install Cloud Pak for Data with selective modules.

Create OCP project for individual modules

oc new-project zen
oc project zen # Ignore the warning message if you get one

Install Cloud Pak for Data Lite

cd /nfs/cpd
./cpd-cli adm --assembly lite --namespace zen --repo ./repo.yaml --apply --accept-all-licenses
./cpd-cli install --assembly lite --namespace zen --repo ./repo.yaml --storageclass nfs-client --transfer-image-to=$(oc registry info)/zen --target-registry-username=$(oc whoami) --target-registry-password=$(oc whoami -t) --insecure-skip-tls-verify --cluster-pull-prefix=image-registry.openshift-image-registry.svc:5000/zen --latest-dependency --accept-all-licenses

List available patches

cd /nfs/cpd
./cpd-cli status -a lite -n zen -r ./repo.yaml --patches --available-updates

Apply patch

Find the latest patch applicable to the Lite assembly and apply.

PATCH_NAME="patch_name"
./cpd-cli patch -a lite -n zen --patch-name $PATCH_NAME --transfer-image-to=$(oc registry info)/zen -r ./repo.yaml --target-registry-username=$(oc whoami) --target-registry-password=$(oc whoami -t) --insecure-skip-tls-verify --cluster-pull-prefix=image-registry.openshift-image-registry.svc:5000/zen

Access Cloud Pak for Data

Once the installation of the Lite assembly has finished, you can access Cloud Pak for Data via the following URL: https://zen-cpd-zen.apps.ocp45.uk.ibm.com. To ensure your computer can resolve the host name, add an entry to the /etc/hosts file:

<bastion IP> zen-cpd-zen.apps.ocp45.uk.ibm.com

Install Watson Studio

assembly="wsl"
cd /nfs/cpd
./cpd-cli adm -a $assembly --repo ./repo.yaml --namespace zen --apply --accept-all-licenses
./cpd-cli install -a $assembly -n zen -c nfs-client --transfer-image-to=$(oc registry info)/zen -r ./repo.yaml --target-registry-username=$(oc whoami) --target-registry-password=$(oc whoami -t) --insecure-skip-tls-verify --cluster-pull-prefix=image-registry.openshift-image-registry.svc:5000/zen --accept-all-licenses

List available patches

./cpd-cli status -a $assembly -n zen -r ./repo.yaml --patches --available-updates

Apply patch

Find the latest patch applicable to the WSL assembly and apply.

PATCH_NAME="patch_name"
./cpd-cli patch -a wsl -n zen --patch-name $PATCH_NAME --transfer-image-to=$(oc registry info)/zen -r ./repo.yaml --target-registry-username=$(oc whoami) --target-registry-password=$(oc whoami -t) --insecure-skip-tls-verify --cluster-pull-prefix=image-registry.openshift-image-registry.svc:5000/zen

Install other assemblies

You can repeat the above steps for the other assemblies, each time by selecting a different assembly name, for example:

assembly="wml"
...

Watson Machine Learning: wml
Watson Knowledge Catalog: wkc
Data Virtualization: dv
Db2 Warehouse: db2wh
Db2 Event Store: db2eventstore
SPSS Modeler: spss-modeler
Decision Optimization: dods
Cognos Analytics: ca
DataStage: ds

If you want to install an assembly that is not listed above, find the installation instructions here: https://www.ibm.com/support/producthub/icpdata/docs/content/SSQNUZ_latest/svc-nav/head/services.html.

Some assemblies like Watson Discovery require a different storage class or other overrides which can be specified with an override. Please refer to the installation instructions of the specific service for more information.

Install multiple assemblies in one go

If you want to install multiple assemblies in a single go, you can execute in a for-loop. Please ensure that the assemblies do not require special overrides such as Spark.

assemblies="lite wsl wml wkc spss dods rstudio dv"
cd /nfs/cpd
for assembly in $assemblies;do
  echo $assembly
  ./cpd-cli adm -a $assembly --repo ./repo.yaml --namespace zen --apply --accept-all-licenses
  ./cpd-cli install -a $assembly -n zen -c nfs-client --transfer-image-to=$(oc registry info)/zen -r ./repo.yaml --target-registry-username=$(oc whoami) --target-registry-password=$(oc whoami -t) --insecure-skip-tls-verify --cluster-pull-prefix=image-registry.openshift-image-registry.svc:5000/zen --cluster-pull-username=$(oc whoami) --cluster-pull-password=$(oc whoami -t) --accept-all-licenses
done

Prepare installation of Db2 Event Store

If you intend to install Db2 Event Store on the cluster, you need to allocate 3 dedicated nodes for this. Preparation instructions can be found here: Db2 Event Store preparation.

Air-gapped install of Cloud Pak for Data

Unpack the downloaded tar file

tar xzf /tmp/cp4d_downloads.tar.gz -C /