Contact Us 1-800-596-4880
  1. Homepage
  2. Connector Builder

  3. Troubleshooting

Troubleshooting Connector Builder

Learn how to troubleshoot issues with Connector Builder.

Access Error Logs

To access error logs in Connector Builder:

  1. Click View > Output.

  2. In the dropdown menu, select Connector Builder.

To modify the output log level, in the Connector Builder extension, go to Settings and find Linkweave › Logging: Level.

Connector Generation Using an API Spec

Connector Builder supports only OAS 3.0 API specifications. If you encounter issues, you can use Anypoint Code Builder - API Extension to validate the API Spec.

Connector Generation Using AI Assistance

Even when given well-crafted prompts, AI might sometimes hallucinate. This can be influenced by factors such as the specific model in use and the quality of the API documentation it processes.

When creating a project using AI assistance, it’s important to recognize that the initial output is a starting point rather than a finalized solution. You should thoroughly review and validate the AI-generated results, as AI might occasionally misinterpret, omit, or misunderstand aspects of a service’s documentation.

If you encounter issues with a connector generated using AI assistance, ensure you review and compare the AI-generated API specification against the official service documentation. This review should focus on, but not be limited to, these areas:

  • Endpoints: Ensure all required API endpoints are present and accurately defined.

  • HTTP Methods: Verify that each endpoint uses the correct HTTP methods (GET, POST, PUT, DELETE, and so on.).

  • Parameters: Confirm that all necessary parameters, including their data types, formats, locations (path, query, header, body), and validation rules, are correctly specified.

  • Request and Response Schemas: Check that the structures and data types of both request and response bodies are correct, including complex data types like nested objects, arrays, and enums.

  • Authentication and Authorization: Make sure the specification accurately describes authentication and authorization mechanisms (such as API keys or OAuth 2.0) required by the service.

  • Error Handling: Ensure that typical error responses and their corresponding status codes are properly documented.

Unable to Locate Operation for TestConnection Element

This error typically occurs during the initial generation of a connector. To ensure a successful connection test, the system requires a simple GET operation without any required parameters. This operation automatically gets included by the code generator. If no such operation is found in the API specification, you will encounter the Could not find an operation for the TestConnection element error.

If you encounter this error, you have two options:

  • Skip and continue generating the connector: This allows you to proceed with connector generation, but you will need to manually define a custom operation for testing the connection.

  • Cancel generation and modify the API specification: You can halt the generation process, update your API specification to include a GET operation that meets the criteria (no required parameters), and then restart the connector generation.

Deletion of Custom Operations and Types when Regenerating a Connector

To safeguard any custom operations or types you’ve added to prevent their accidental deletion, organize your custom elements within dedicated folders:

  • Custom Operations: Create a customizations folder inside the operations folder and place all your custom operation definitions within it.

  • Custom Types: Similarly, for custom types, create a customizations folder within the types folder and add your custom type definitions there.

By following this structure, the regeneration process recognizes these dedicated folders and preserves your custom work.

Operations Missing from a Connector

Some operations might be missing from a connector if they were removed as part of our mitigation strategies to ensure out-of-the-box compatibility with the target platform.

Details of these removals can be found in the mitigation report that appears upon connector generation or regeneration.

If you want to turn off these mitigations, go to Settings and disable Linkweave › Features › Mitigations.

Parameters Defined as Union Type

In Anypoint Studio, the parameter doesn’t appear in the operation’s UI. In Anypoint Code Builder, the parameter is shown but produces an error when you try to view its details. An example of an unsupported type is type unionParameter = Type1 | Type2. This is because Anypoint Connectors don’t allow for any of the parameters to be defined as a union type.

Anypoint Connector operations are defined as transformations of the canonical connectivity model (CCM) operations. For example:

type anypoint_O_projects_api_v3_projects_updates_json_get_request = {
  updatedAfter?: O_projects_api_v3_projects_updates_json_get_Type.request.query.updatedAfter,
  // ... other fields
}

var anypoint_O_projects_api_v3_projects_updates_json_get_mapping = [
  "updatedAfter" mapsTo "query.updatedAfter",
  // ... other mappings
  {} defaultOf "query",
  {} alwaysMapsTo "headers",
  {} alwaysMapsTo "cookie"
]

@OperationElement()
var anypoint_O_projects_api_v3_projects_updates_json_get = O_projects_api_v3_projects_updates_json_get withTransformer {
  in: fromMapping<anypoint_O_projects_api_v3_projects_updates_json_get_request, O_projects_api_v3_projects_updates_json_get_Type.request>(anypoint_O_projects_api_v3_projects_updates_json_get_mapping),
  out: extractRequestBody
}

This code translates into an Anypoint Connector operation with all fields as parameters of the CCM operation. None of the field values can be of union type. In the example, O_projects_api_v3_projects_updates_json_get_Type.request.query.updatedAfter can’t be a union type. To determine this manually, you must navigate to the definition of this type in the connectivity model to see if it is a union type.

You can either:

  • Omit the operation.

  • Remove the parameter from the operation.

  • Change the parameter’s type to Object or another appropriate type.

LLM Asks for Input

If the LLM asks for input, provide the input or restart the conversation in a new chat window.

LLM Doesn’t Create Files

If the LLM doesn’t create the files, ensure that you approve the changes proposed by the LLM, both for discovery.json and the API specification.

LLM Indicates Completion, but Files Are Missing

Occasionally, the LLM might incorrectly state that it has created the API specification when it has not. In such cases, you can manually copy the API specification content provided by the LLM, save it as a YAML file in the project root, and the extension should automatically detect the file and begin the generation process.

See Also