Removing a Data Gate for watsonx instance

Edit online

Delete a Data Gate for watsonx instance that you no longer need, or delete an instance to create a new instance with the same name.

About this task

A removal of an instance from the Cloud Pak for Data user interface sometimes does not delete all related objects of that instance. That is, you might be unable to pair a new instance with the same route host and port to the same Db2® for z/OS® subsystem or data sharing group without doing some clean-up work before.

Procedure

  1. Make sure that the Cloud Pak for Data perspective is selected. If necessary, click Icon (button) to change the perspective to change the perspective.
  2. On the side bar, click Instances.
    You see the Service instances page. All instances are listed in the Name column.
  3. Locate and click the menu of the instance that you want to work on.
  4. From that drop-down menu (Drop-down menu on instances table on My instances page.), select Delete.
    A confirmation window is displayed:
    Figure 1. Instance-removal confirmation window
    Confirmation window that is displayed after you select the "Delete" option for an instance
  5. Click Delete.

Results

After completing the steps in the procedure, your instance should be removed entirely. However, sometimes this does not work properly, and your instance is still listed with a status of deleting, which seems to be displayed endlessly. If so, proceed as follows:
  1. Look at the address bar of the browser window in which your instance is displayed. Make a note of the instance ID. The address bar shows a URL string that might look like /data-gate/#/details/dg-1719836825931627, in which case 1719836825931627 is the instance ID.
  2. Open a terminal session with the OpenShift® cluster where Data Gate for watsonx is installed.
  3. In the terminal window, type the following command at the command prompt. In the command, replace <your-instance-id> with the ID you noted as you completed step 1.
    kubectl patch datagateinstanceservice dg<your-instance-id> 
   --patch '[{"op":"remove","path":"/metadata/finalizers"}]' 
   --type json

What to do next

The removal function does not remove all related objects. This might surface as a problem in the following situations:

  • Db2 for z/OS data sources only: You want to pair a new instance with the same route host and port to the same Db2 for z/OS subsystem or data sharing group. In this case, the removal function cannot remove all related objects from the Db2 for z/OS communications database.

    Therefore, you must clean up the communications database manually. Follow Cleaning up the Db2 for z/OS communications database for instructions.

  • You want to pair a new instance with the same route host and port, and want to use the same storage as the removed instance (existing storage).

    In this case, remove the following folders and files manually from the storage entity:

    • appliance_etc
    • clone-api
    • dwa
    • dwa-apply-container-start.log
    • scratch
    • stunnel.conf
    • stunnel.pem
    • stunnel.pem.old

The Parquet objects that are created for the replication of VSAM or IMS data in the connected watsonx.data bucket are not removed automatically by dropping the table. This does not lead to problems, but you might want to free up the bucket space and delete these objects. For this purpose, you can use the ibm-lh tool. See Spark table maintenance by using ibm-lh tool for more information.