Salesforce MuleSoft Mule-Dev-301 Book

For the Mule 4 custom-policy development model represented by this question, the policy's runtime behavior is implemented in an XML template, while the policy's configurable parameters and metadata are described in a YAML configuration file.

The XML template uses Mule's XML configuration language and policy-specific elements such as http-policy:proxy, http-policy:source, http-policy:operation, and http-policy:execute-next. In this scenario, the developer can place the header-token inspection and validation behavior in this implementation layer. The accompanying YAML file exposes configurable policy parameters to API Manager and provides the metadata needed for policy configuration.

MuleSoft's custom-policy development reference identifies an XML template and YAML file as core policy project artifacts; the complete project additionally contains its POM, Mule artifact descriptor, and any optional resources.

A ZIP file is not the source artifact pair used to define the policy. Similarly, a generic JSON properties file is not a replacement for the implementation template in this Mule 4 custom-policy model.

Thus, among the available combinations, the XML policy template together with the YAML configuration file is the required pair.

Reference topics: Mule Gateway custom policies; XML policy templates; YAML policy configuration; HTTP Policy Transform Extension.

Official documentation:

===============================================================

Anypoint Monitoring provides connector-level performance metrics without requiring additional application logging or custom instrumentation. For supported connectors, these statistics are accessible from the application's monitoring interface through the Connectors tab.

MuleSoft's documentation specifies that connector monitoring includes Requests, which reports the average number of connector requests over automatically selected intervals, and Response Time, which reports the average response time for connector requests. To access these metrics, the documented procedure is to open Anypoint Monitoring, select the relevant Mule application resource, and open the Connectors view.

Because these metrics are generated by the monitoring platform, examining raw logs is unnecessary. A custom dashboard may be useful when creating specialized operational visualizations, but it is not required simply to obtain standard connector request and response-time measurements. Modifying the Salesforce System API to manually log this information would also duplicate functionality already supplied by Anypoint Monitoring and increase application complexity.

The question additionally identifies Titanium support, which aligns with availability of connector monitoring capabilities described by MuleSoft.

Therefore, the correct procedure is to use the built-in Anypoint Monitoring dashboard and inspect the Connectors tab.

Reference topics: Anypoint Monitoring; connector performance metrics; built-in dashboards; request count; response time; Salesforce Connector.

Official documentation:

===============================================================

The significant attribute in the MUnit definition is expectedErrorType="APP:CUSTOM_ERROR". The referenced implementation flow deliberately invokes raise-error with exactly the same error type: APP:CUSTOM_ERROR.

MUnit supports expected-error testing specifically for negative scenarios. When expectedErrorType is configured, the test is considered successful when execution produces that expected Mule error. The raised error is therefore not treated as an unexpected test failure; it satisfies the test's declared expectation.

The apparently failing assertion—testing true against equalTo(false)—does not change the result in this scenario because normal execution does not reach the validation chain after the referenced flow raises the expected error. The expected error terminates the execution path being tested and satisfies the MUnit expectation.

If the implementation generated a different error type, option B would apply. If it generated no error when an error was expected, the test would also fail. Here, however, the flow explicitly raises the exact application-defined error specified by the test.

Reference topics: MUnit Test Structure; expectedErrorType; Error Testing; Raise Error Component.

Official documentation:

===============================================================

When any route within Scatter-Gather fails, Mule raises a MULE:COMPOSITE_ROUTING error after the routes complete. The information associated with each route is retained in the error message payload. This payload contains separate collections for failures and results, indexed according to the route positions.

Routes are zero-indexed for this structure: Route A corresponds to index 0, Route B to index 1, Route C to index 2, and Route D to index 3. Because Route C failed, its failure information is therefore available from the failures structure at index 2.

Critically, this data belongs to the Mule error's errorMessage.payload; it is not simply the current event payload. Consequently, payload.failures[2] and payload[2] address the wrong structure. Likewise, error.errorMessage.payload.results[2] refers to the successful-results collection; for a failed route, the relevant object is stored under failures.

MuleSoft's Scatter-Gather documentation demonstrates the same mechanism, accessing failed route information through expressions such as error.errorMessage.payload.failures['0'] and successful route output through the corresponding results collection.

Reference topics: Scatter-Gather Error Handling; MULE:COMPOSITE_ROUTING; Route Results and Failures; Compensation Logic.

Official documentation:

===============================================================