Micronaut：@Endpoint注解id的配置占位符忽略自定义值问题

Great question! This usage is indeed supported by Spring Boot Actuator, but the issue you're facing stems from the timing of how Spring resolves the @Endpoint annotation's id property. Let's break down the root causes and fixes:

Why the Custom Value Isn't Being Applied #

The @Endpoint annotation's attributes are resolved during the Bean Definition phase of the Spring application context lifecycle—this happens very early, before the full application context is refreshed. If your custom jupyter.endpoint.path property isn't available to Spring's Environment at this early stage, the default value will be used instead.

Common Causes #

1. Your custom property isn't properly loaded into Spring's Environment

   • Double-check that your property (e.g., jupyter.endpoint.path=command/jupyter/start-kernel) is actually being read by Spring. You can verify this by injecting Environment into a component and printing the value:

     @Component
     public class ConfigChecker {
         public ConfigChecker(Environment env) {
             System.out.println("Endpoint path property: " + env.getProperty("jupyter.endpoint.path"));
         }
     }
     

   • Ensure your configuration file (application.properties/yaml) is in the correct location (usually src/main/resources), or if using external config, pass it via startup arguments (--jupyter.endpoint.path=xxx) or environment variables (JUPYTER_ENDPOINT_PATH=xxx).

2. Custom property sources are added too late

   • If you're loading the property via a custom PropertySource that's registered after the Bean Definition phase (e.g., in a @PostConstruct method), Spring won't see it when resolving the @Endpoint id.

Fixes to Try #

1. Ensure the Property is Loaded Early #

Use one of these methods to make sure the property is available during the Bean Definition phase:

• Pass the property as a command-line argument when starting the app:

  java -jar your-app.jar --jupyter.endpoint.path=command/jupyter/start-kernel
  

• Set it as an environment variable (Spring converts snake_case to kebab-case automatically):

  export JUPYTER_ENDPOINT_PATH=command/jupyter/start-kernel
  java -jar your-app.jar
  

2. Register the Endpoint Programmatically #

If you need dynamic control over the endpoint ID at runtime, skip the @Endpoint annotation and register the endpoint programmatically. This lets you resolve the property directly from the Environment when registering:

@Configuration
public class KernelEndpointConfig {

    @Bean
    public KernelEndpoint kernelEndpoint() {
        return new KernelEndpoint();
    }

    @Bean
    public EndpointRegistration kernelEndpointRegistration(KernelEndpoint endpoint, Environment environment) {
        // Resolve the property with default fallback
        String endpointId = environment.getProperty("jupyter.endpoint.path", "jupyter/kernel");
        // Register the endpoint with the resolved ID
        return EndpointRegistration.of(endpointId, endpoint)
                .withWriteOperations("start"); // Map your @Write method here
    }
}

// Modify your KernelEndpoint to remove the @Endpoint annotation
public class KernelEndpoint {
    public Map<String, String> start() {
        return Map.of("message", "Kernel start request received!");
    }
}

3. Verify Spring Boot Actuator Version #

Ensure you're using a Spring Boot version that supports SpEL in @Endpoint annotations (all Spring Boot 2.x versions do, but very old 1.x versions don't). If you're on an outdated version, upgrade to a stable release.

内容的提问来源于stack exchange，提问作者JoshuaCWebDeveloper