Known limitations
For the most up-to-date information, see the support page IBM Automation Decision Services Known Limitations, which is regularly updated.
Setting up your Automation Decision Services environment
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Installing Automation Decision Services | The embedded MongoDB database is not highly available (HA). | The IT user can provide a remote MongoDB database, and set the MongoDB connection string URI to the remote MongoDB. For more information about configuring the MongoDB secret, see Configuring MongoDB storage. |
| The embedded MongoDB persistent volume requires a fully POSIX-compliant storage. | Make sure to use a storage class that is fully POSIX-compliant. An azure-file storage, for example, does not meet this requirement. | |
| Using the sample CI/CD stack | Some container images run as root user. | None |
Administering
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Connecting to a remote repository automatically | You cannot create a global Git provider by using the SSH method to connect to a remote repository automatically. | None |
| Connecting to a remote repository |
You cannot use the force push command ( |
Disable the force push feature directly on the remote repository; for example, through the GitHub interface. If you used the force push command, you can use one of the two following ways to repair: Option 1: Restore a backup of the remote repository. Some data might be lost depending on the backup strategy. Option 2:
|
Managing decisions
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Importing decision services | Decision service archives that were compressed using the default compression tool on your computer cannot be opened in Decision Designer. |
Import decision service archives created with the Export feature. For more information, see Managing decision services. |
| When you import a decision automation that was initially created and exported from the same Decision Designer instance you are working in, the predictive models it contains will run even if they are not connected to a machine learning provider. | The provider details are found in the original decision automation settings at runtime. | |
| Sharing changes | When you share a large decision service, an error might occur (HTTP response code 500). |
The error is thrown when the operation has not completed on time. Check the Share changes tab after 2 minutes. If there are still pending changes, contact your administrator. To avoid running into this error, try to share your modifications often when you build a large decision service. |
| You cannot upload a file for a decision service that is larger than 64 MB to a Git server. | Configure the maximum file size upload by setting the spec.decision_designer.git_service.max_upload_file_size parameter in the git-service. | |
| Upgrading decision services | Upgrading a decision service fails when both of the following conditions are met:
|
Include the missing external libraries in the decision automation before the upgrade. For more information about using external libraries, see Importing an external library into. |
| Creating branches | You cannot create a new branch unless your decision automation is connected to a remote Git repository (remote repository). | Connect to the remote repository before you create a new branch. For more information, see Connecting to a remote repository automatically and Connecting to a remote repository manually. |
Modeling decisions
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Building an external library | When you build an external library, no warning is displayed if:
|
If the methods that you want to use are a pure function, you can declare them as such by adding a Java or external annotation. For more information, see Pure functions. |
| Importing an external library into Decision Designer | If your decision service contains an external library with a different locale than the decision service, an error occurs in the data model and no data type is available. |
|
| Creating a decision model | If the decision is multi-valued or if you selected the first rule applies policy, you cannot add a decision node with the same name as an attribute that is used in its decision logic. | Choose a different name for the decision node, or set a different output variable name. |
| Creating a task model | If there are more than 3000 rows in a decision table, it can increase editing and build time, resulting in performance degradation. | Reduce the number of rows in the decision table. If possible, break up larger decision tables into smaller tables. |
| If a rule task node contains more than 5000 actions, it can temporarily block the Java virtual machine, resulting in considerable performance degradation. | Reduce the number of actions in the rule task node. | |
| Adding the decision logic | When you use the copy of ... where ... statement in a rule, an error occurs
if one of the specified values must be converted (for example, from list to array). |
None |
| Static attributes from external libraries can be modified without restrictions, even though they should not. | None | |
| Composing and reusing models | When you update a decision model that is used as a function in another decision model, the changes are not reflected in the decision model where it is integrated. | Reset the type of the function node. |
| When you update a task model that is used as a function in another task model, the changes are not reflected in the task model where it is integrated. | Remove the function task node and add it again. | |
| Importing data models | After upgrading from 24.0.0 to 24.0.0-IF003, a warning is displayed if JSON schemas defining the same data type with int32 attributes are imported both before and after the upgrade.
This issue arises only if int32 attributes were imported before the upgrade, as they are converted to int64, causing a mismatch when the same attributes are reused after the iFix. |
To resolve the data type duplication warning, delete and reimport the JSON schemas that were imported before upgrading to 24.0.0-IF003. |
Integrating machine learning
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Managing machine learning providers | Watson Machine Learning functions cannot be used in Automation Decision Services. | None |
| You cannot connect to Watson Machine Learning that is hosted on IBM Cloud Pak for Data. | None |
Building and deploying decision services
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Building and deploying in a CI/CD stack | You cannot run unit tests in a CI/CD stack on predictive models or other models that call predictive models because credentials cannot be passed. | None |
| Building and deploying from Decision Designer | Custom POM files are ignored in the embedded build service. | None |
| You cannot run unit tests in Decision Designer | None | |
| When you upgraded your decision services from 23.0.2 to 24.0.0, the upgraded decision services are not listed in the Deploy tab in Decision Designer. |
Although the upgraded decision services are not listed in Decision Designer, you can still execute them. You can find the decision services by using the following endpoint in Swagger
UI:
|
Executing decision services
| Feature | Limitation | Comment or workaround |
|---|---|---|
| Calling decision services | Users with the decision service manager and decision runtime
monitor roles are also granted the decision service user role and can
execute decision services. |
None |