Update 1

This use case demonstrates dynamically adding a data source to your running domain by updating your model and rolling your domain. It demonstrates several features of WDT and Model in Image:

  • The syntax used for updating a model is the same syntax you use for creating the original model.
  • A domain’s model can be updated dynamically by supplying a model update in a file in a Kubernetes ConfigMap.
  • Model updates can be as simple as changing the value of a single attribute, or more complex, such as adding a JMS Server.

For a detailed description of model updates, see Runtime Updates in the Model in Image user guide.

The operator does not support all possible dynamic model updates. For model update limitations, consult Runtime Updates in the Model in Image user docs, and carefully test any model update before attempting a dynamic update in production.

Here are the steps:

  1. Ensure that you have a running domain.

    Make sure you have deployed the domain from the Initial use case.

  2. Create a data source model YAML file.

    Create a WDT model snippet for a data source (or use the example provided). Make sure that its target is set to cluster-1, and that its initial capacity is set to 0.

    The reason for the latter is to prevent the data source from causing a WebLogic Server startup failure if it can’t find the database, which would be likely to happen because you haven’t deployed one (unless you’re using the JRF path through the sample).

    Here’s an example data source model configuration that meets these criteria:

    resources:
  JDBCSystemResource:
    mynewdatasource:
      Target: 'cluster-1'
      JdbcResource:
        JDBCDataSourceParams:
          JNDIName: [
            jdbc/mydatasource1,
            jdbc/mydatasource2
          ]
          GlobalTransactionsProtocol: TwoPhaseCommit
        JDBCDriverParams:
          DriverName: oracle.jdbc.xa.client.OracleXADataSource
          URL: '@@SECRET:@@ENV:DOMAIN_UID@@-datasource-secret:url@@'
          PasswordEncrypted: '@@SECRET:@@ENV:DOMAIN_UID@@-datasource-secret:password@@'
          Properties:
            user:
              Value: '@@SECRET:@@ENV:DOMAIN_UID@@-datasource-secret:user@@'
            oracle.net.CONNECT_TIMEOUT:
              Value: 5000
            oracle.jdbc.ReadTimeout:
              Value: 30000
        JDBCConnectionPoolParams:
            InitialCapacity: 0
            MaxCapacity: '@@SECRET:@@ENV:DOMAIN_UID@@-datasource-secret:max-capacity@@'
            TestTableName: SQL ISVALID
            TestConnectionsOnReserve: true

    Place the above model snippet in a file named /tmp/mii-sample/mydatasource.yaml and then use it in the later step where you deploy the model ConfigMap, or alternatively, use the same data source that’s provided in /tmp/mii-sample/model-configmaps/datasource/model.20.datasource.yaml.

  3. Create the data source secret.

    The data source references a new secret that needs to be created. Run the following commands to create the secret:

    $ kubectl -n sample-domain1-ns create secret generic \
   sample-domain1-datasource-secret \
   --from-literal='user=sys as sysdba' \
   --from-literal='password=incorrect_password' \
   --from-literal='max-capacity=1' \
   --from-literal='url=jdbc:oracle:thin:@oracle-db.default.svc.cluster.local:1521/devpdb.k8s'

    $ kubectl -n sample-domain1-ns label  secret \
   sample-domain1-datasource-secret \
   weblogic.domainUID=sample-domain1

    We deliberately specify an incorrect password and a low maximum pool capacity because we will demonstrate dynamically correcting the data source attributes in the Update 4 use case without requiring rolling the domain.

    You name and label secrets using their associated domain UID for two reasons:

    • To make it obvious which secret belongs to which domains.
    • To make it easier to clean up a domain. Typical cleanup scripts use the weblogic.domainUID label as a convenience for finding all the resources associated with a domain.

  4. Create a ConfigMap with the WDT model that contains the data source definition.

    Run the following commands:

    $ kubectl -n sample-domain1-ns create configmap sample-domain1-wdt-config-map \
  --from-file=/tmp/mii-sample/model-configmaps/datasource

    $ kubectl -n sample-domain1-ns label configmap sample-domain1-wdt-config-map \
  weblogic.domainUID=sample-domain1
    • If you’ve created your own data source file, then substitute the file name in the --from-file= parameter (we suggested /tmp/mii-sample/mydatasource.yaml earlier).
    • Note that the -from-file= parameter can reference a single file, in which case it puts the designated file in the ConfigMap, or it can reference a directory, in which case it populates the ConfigMap with all of the files in the designated directory.

    You name and label the ConfigMap using its associated domain UID for two reasons:

    • To make it obvious which ConfigMap belong to which domains.
    • To make it easier to clean up a domain. Typical cleanup scripts use the weblogic.domainUID label as a convenience for finding all resources associated with a domain.

  5. Update your Domain YAML file to refer to the ConfigMap and Secret.

    • Option 1: Update a copy of your Domain YAML file from the Initial use case.

      • In the Initial use case, we suggested creating a Domain YAML file named /tmp/mii-sample/mii-initial.yaml or using the /tmp/mii-sample/domain-resources/WLS/mii-initial-d1-WLS-v1.yaml file that is supplied with the sample.

        • We suggest copying the original Domain YAML file and naming the copy /tmp/mii-sample/mii-update1.yaml before making any changes.

        • Working on a copy is not strictly necessary, but it helps keep track of your work for the different use cases in this sample and provides you a backup of your previous work.

      • Add the secret to its spec.configuration.secrets stanza:

        spec:
  ...
  configuration:
  ...
    secrets:
    - sample-domain1-datasource-secret

        (Leave any existing secrets in place.)

      • Change its spec.configuration.model.configMap to look like the following:

        spec:
  ...
  configuration:
    ...
    model:
      ...
      configMap: sample-domain1-wdt-config-map

      • Apply your changed Domain YAML file:

        NOTE: Before you deploy the domain custom resource, determine if you have Kubernetes cluster worker nodes that are remote to your local machine. If so, then you need to put the Domain YAML file’s image in a location that these nodes can access and you may also need to modify your Domain YAML file to reference the new location. See Ensuring your Kubernetes cluster can access images.

        $ kubectl apply -f /tmp/mii-sample/mii-update1.yaml

    • Option 2: Use the updated Domain YAML file that is supplied with the sample:

      NOTE: Before you deploy the domain custom resource, determine if you have Kubernetes cluster worker nodes that are remote to your local machine. If so, then you need to put the Domain YAML file’s image in a location that these nodes can access and you may also need to modify your Domain YAML file to reference the new location. See Ensuring your Kubernetes cluster can access images.

      $ kubectl apply -f /tmp/miisample/domain-resources/WLS/mii-update1-d1-WLS-v1-ds.yaml

  6. Restart (‘roll’) the domain.

    Now that the data source is deployed in a ConfigMap and its secret is also deployed, and you have applied an updated Domain YAML file with its spec.configuration.model.configMap and spec.configuration.secrets referencing the ConfigMap and secret, tell the operator to roll the domain.

    When a model domain restarts, it will rerun its introspector job to regenerate its configuration, and it will also pass the configuration changes found by the introspector to each restarted server. One way to cause a running domain to restart is to change the domain’s spec.restartVersion. To do this:

    • Option 1: Edit your domain custom resource.

      • Call kubectl -n sample-domain1-ns edit domain sample-domain1.
      • Edit the value of the spec.restartVersion field and save.
        • The field is a string; typically, you use a number in this field and increment it with each restart.

    • Option 2: Dynamically change your domain using kubectl patch.

      • To get the current restartVersion call:

        $ kubectl -n sample-domain1-ns get domain sample-domain1 '-o=jsonpath={.spec.restartVersion}'

      • Choose a new restart version that’s different from the current restart version.

        • The field is a string; typically, you use a number in this field and increment it with each restart.

      • Use kubectl patch to set the new value. For example, assuming the new restart version is 2:

        $ kubectl -n sample-domain1-ns patch domain sample-domain1 --type=json '-p=[{"op": "replace", "path": "/spec/restartVersion", "value": "2" }]'

    • Option 3: Use the sample helper script.

      • Call /tmp/mii-sample/utils/patch-restart-version.sh -n sample-domain1-ns -d sample-domain1.
      • This will perform the same kubectl get and kubectl patch commands as Option 2.

  7. Wait for the roll to complete.

    Now that you’ve started a domain roll, you’ll need to wait for it to complete if you want to verify that the data source was deployed.

    • One way to do this is to call kubectl get pods -n sample-domain1-ns --watch and wait for the pods to cycle back to their ready state.

    • Alternatively, you can run /tmp/mii-sample/utils/wl-pod-wait.sh -p 3. This is a utility script that provides useful information about a domain’s pods and waits for them to reach a ready state, reach their target restartVersion, and reach their target image before exiting.

      Click here to display the `wl-pod-wait.sh` usage.
      Click here to view sample output from `wl-pod-wait.sh` that shows a rolling domain.

  8. After your domain is running, you can call the sample web application to determine if the data source was deployed.

    Send a web application request to the ingress controller:

    $ curl -s -S -m 10 -H 'host: sample-domain1-cluster-cluster-1.mii-sample.org' \
   http://localhost:30305/myapp_war/index.jsp

    Or, if Traefik is unavailable and your Administration Server pod is running, you can run kubectl exec:

    $ kubectl exec -n sample-domain1-ns sample-domain1-admin-server -- bash -c \
  "curl -s -S -m 10 http://sample-domain1-cluster-cluster-1:8001/myapp_war/index.jsp"

    You will see something like the following:

    <html><body><pre>
*****************************************************************

Hello World! This is version 'v1' of the mii-sample JSP web-app.

Welcome to WebLogic Server 'managed-server1'!

  domain UID  = 'sample-domain1'
  domain name = 'domain1'

Found 1 local cluster runtime:
  Cluster 'cluster-1'

Found min threads constraint runtime named 'SampleMinThreads' with configured count: 1

Found max threads constraint runtime named 'SampleMaxThreads' with configured count: 10

Found 1 local data source:
  Datasource 'mynewdatasource':  State='Running', testPool='Failed'
    ---TestPool Failure Reason---
    NOTE: Ignore 'mynewdatasource' failures until the sample's Update 4 use case.
    ---
    ...
    ... invalid host/username/password
    ...
    -----------------------------

*****************************************************************
</pre></body></html>

A TestPool Failure is expected because we will demonstrate dynamically correcting the data source attributes in Update 4.

If you see an error other than the expected TestPool Failure, then consult Debugging in the Model in Image user guide.

If you plan to run the Update 3 or Update 4 use case, then leave your domain running.

To remove the resources you have created in the samples, see Cleanup.