Table of contents

Manually resuming installations from a specific module

If your Cloud Pak for Data installation fails, you can manually resume the installation from a specific module.

Before you begin

Required role: To complete this task, you must be a cluster administrator.

About this task

The following resume process assumes the module information (including chart name, architecture and version) remains the same during the resume operation. This operation enables you to trigger the resume from a module that has failed during installation. Ensure you fix the reasons for the installation failure before you attempt the resume.

Procedure

  1. Log in to the cluster as an administrator:
    oc login OpenShift_URL:port
  2. Change to the project that you are installing Cloud Pak for Data in:
    oc project <project name>
  3. Go to the operator pod to use the helm command. You might need to enter oc get po to get the right pod name.
    oc exec -it cpd-install-operator-xxxxxx -- /bin/sh
  4. In the pod terminal, specify the tiller environment variables:
    export HELM_HOST=tiller-svc:44134
    export HELM_TLS_CA_CERT=/etc/certs/ca.cert.pem
    export HELM_TLS_CERT=/etc/certs/helm.cert.pem
    export HELM_TLS_KEY=/etc/certs/helm.key.pem
  5. Ensure helm and tiller are working. Enter the following command to return the helm version:
    helm version --tls
  6. Delete the release using helm. To list the release name, run helm list --tls first.
    helm delete <release name> --purge --tls
  7. Exit the pod terminal:
    exit
  8. Delete all the pvc and pv related to the module:
    oc delete pvc <pvc name>
    oc delete pv <pv name>
  9. Open cpd-install-spec in edit mode:
    oc edit cm cpd-install-spec
  10. In the spec.modules section, update the retryCount: field.
    Updating cpd-install-spec automatically re-triggers the installation.
  11. You can check the operator pod log to get installation progress information:
    oc logs $(oc get po | grep cpd-install-operator | awk '{print $1}')