Connecting Watson Machine Learning Accelerator to Watson Machine Learning

After deploying both the Watson Machine Learning service and the Watson Machine Learning Accelerator service, you can connect the two services to share training resources and use Watson Machine Learning resources with Watson Machine Learning Accelerator.

Procedure

  1. If Watson Machine Learning and Watson Machine Learning Accelerator are installed in different Cloud Pak for Data namespaces.
    1. Get the JWT public key used by the Watson Machine Learning service deployment which can be found on the WML training pod in the /user-home/public.pem location:
      kubectl -n wml-namespace exec -it wmltraining bash
      cat /user-home/public.pem
      where wml-namespace is the WML namespace and wmltraining is the name of the WML training pod. The WML training pod name starts with wmltraining-*, for example: wmltraining-6b648cdd4-x9ksh

    2. Update the public.pem key value in the dlpd pod of WML Accelerator.
      1. Get the dlpd pod name from the WML Accelerator namespace.
        oc get pods -n wmla-tether-namespace
        where wmla-tether-namespace is the name of the WML Accelerator deployed namespace.

        The dlpd pod name starts with wmla-dlpd-*, for example: wmla-dlpd-dcbdb6bc7-m8k8p.

      2. SSH into the dlpd service pod to get the JWT token information.
        oc exec -it wmla-dlpd-dcbdb6bc7-m8k8p -c dlpd sh -n wmla-tether-namespace
      3. Get the current JWT public key set in the dlpd service using the below commands:
        cat /var/shareDir/dli/conf/dlpd/dlpd.conf | grep -i JWT
        "DLI_JWT_SECRET_KEY" : "/var/shareDir/dli/conf/dlpd/cpd_jwt.pub",
        "DLI_JWT_VERIFY_NOTBEFORE_TIME": "on",
        "DLI_JWT_VERIFY_EXPIRATION_TIME": "on",
        "DLI_JWT_SINGLE_USER_MODE": "on",
        "DLI_JWT_SINGLE_USER_NAME": "Admin",
      4. Update the WML JWT token in cpd_jwt.pub. The location of this file is specified by the DLI_JWT_SECRET_KEY parameter in the previous step, for example: /var/shareDir/dli/conf/dlpd/cpd_jwt.pub.
    3. Restart the dlpd service:
      cd /var/shareDir/dli
      /opt/ibm/spectrumcomputing/dli/2.3.0/dlpd/tools/msd/msdctl restart_dlpd
  2. In WML, run the updateWMLClusterdetails.sh command line utility which allows WML to locate and use a WML Accelerator instance. The script is located in the WML case package.
    1. Obtain and extract the WML files from the case package (ibm-wml-cpd-4.0.0.tgz).
      tar -zxvf ibm-wml-cpd-4.0.0.tgz
    2. Navigate to the ibm-wml-cpd/inventory/wmlOperatorSetup/files directory.
    3. Run the updateWMLClusterdetails.sh script.
      sh updateWMLClusterdetails.sh wmla-dlpd.wmla-namespace.svc.cluster.local 9243 wml-ig wml-ig-edt wml_routename
      where:
      • wmla-namespace is the name of the Watson Machine Learning Accelerator tethered namespace.
      • wml_routename is the WML route name. To obtain the route name, issue the following command:
        oc get route/cpd -o jsonpath={.spec.host} -n cpd-namespace
        where cpd-namespace is the Cloud Pak for Data namespace.
      For example, where the wmla-namespace is wml-accelerator-ns and wml_routename is wml-route:
      sh updateWMLClusterdetails.sh wmla-dlpd.wml-accelerator-ns.svc.cluster.local 9243 wml-ig wml-ig-edt wml-route
  3. Verify that the connection was successful by checking that the Deep Learning experiment option is available.

What to do next