Spring Boot Features

This section dives into the details of Spring Boot. Here you can learn about the key features that you may want to use and customize. If you have not already done so, you might want to read the "getting-started.html" and "using-spring-boot.html" sections, so that you have a good grounding of the basics.

1. SpringApplication

The SpringApplication class provides a convenient way to bootstrap a Spring application that is started from a main() method. In many situations, you can delegate to the static SpringApplication.run method, as shown in the following example:

public static void main(String[] args) {
    SpringApplication.run(MySpringConfiguration.class, args);
}

When your application starts, you should see something similar to the following output:

  .   ____          _            __ _ _
 /\\ / ___'_ __ _ _(_)_ __  __ _ \ \ \ \
( ( )\___ | '_ | '_| | '_ \/ _` | \ \ \ \
 \\/  ___)| |_)| | | | | || (_| |  ) ) ) )
  '  |____| .__|_| |_|_| |_\__, | / / / /
 =========|_|==============|___/=/_/_/_/
 :: Spring Boot ::   v2.4.6

2019-04-31 13:09:54.117  INFO 56603 --- [           main] o.s.b.s.app.SampleApplication            : Starting SampleApplication v0.1.0 on mycomputer with PID 56603 (/apps/myapp.jar started by pwebb)
2019-04-31 13:09:54.166  INFO 56603 --- [           main] ationConfigServletWebServerApplicationContext : Refreshing org.springframework.boot.web.ser[email protected] 6e5a8246: startup date [Wed Jul 31 00:08:16 PDT 2013]; root of context hierarchy
2019-04-01 13:09:56.912  INFO 41370 --- [           main] .t.TomcatServletWebServerFactory : Server initialized with port: 8080
2019-04-01 13:09:57.501  INFO 41370 --- [           main] o.s.b.s.app.SampleApplication            : Started SampleApplication in 2.992 seconds (JVM running for 3.658)

By default, INFO logging messages are shown, including some relevant startup details, such as the user that launched the application. If you need a log level other than INFO, you can set it, as described in Log Levels. The application version is determined using the implementation version from the main application class’s package. Startup information logging can be turned off by setting spring.main.log-startup-info to false. This will also turn off logging of the application’s active profiles.

To add additional logging during startup, you can override logStartupInfo(boolean) in a subclass of SpringApplication.

1.1. Startup Failure

If your application fails to start, registered FailureAnalyzers get a chance to provide a dedicated error message and a concrete action to fix the problem. For instance, if you start a web application on port 8080 and that port is already in use, you should see something similar to the following message:

***************************
APPLICATION FAILED TO START
***************************

Description:

Embedded servlet container failed to start. Port 8080 was already in use.

Action:

Identify and stop the process that's listening on port 8080 or configure this application to listen on another port.
Spring Boot provides numerous FailureAnalyzer implementations, and you can add your own.

If no failure analyzers are able to handle the exception, you can still display the full conditions report to better understand what went wrong. To do so, you need to enable the debug property or enable DEBUG logging for org.springframework.boot.autoconfigure.logging.ConditionEvaluationReportLoggingListener.

For instance, if you are running your application by using java -jar, you can enable the debug property as follows:

$ java -jar myproject-0.0.1-SNAPSHOT.jar --debug

1.2. Lazy Initialization

SpringApplication allows an application to be initialized lazily. When lazy initialization is enabled, beans are created as they are needed rather than during application startup. As a result, enabling lazy initialization can reduce the time that it takes your application to start. In a web application, enabling lazy initialization will result in many web-related beans not being initialized until an HTTP request is received.

A downside of lazy initialization is that it can delay the discovery of a problem with the application. If a misconfigured bean is initialized lazily, a failure will no longer occur during startup and the problem will only become apparent when the bean is initialized. Care must also be taken to ensure that the JVM has sufficient memory to accommodate all of the application’s beans and not just those that are initialized during startup. For these reasons, lazy initialization is not enabled by default and it is recommended that fine-tuning of the JVM’s heap size is done before enabling lazy initialization.

Lazy initialization can be enabled programmatically using the lazyInitialization method on SpringApplicationBuilder or the setLazyInitialization method on SpringApplication. Alternatively, it can be enabled using the spring.main.lazy-initialization property as shown in the following example:

Properties
spring.main.lazy-initialization=true
Yaml
spring:
  main:
    lazy-initialization: true
If you want to disable lazy initialization for certain beans while using lazy initialization for the rest of the application, you can explicitly set their lazy attribute to false using the @Lazy(false) annotation.

1.3. Customizing the Banner

The banner that is printed on start up can be changed by adding a banner.txt file to your classpath or by setting the spring.banner.location property to the location of such a file. If the file has an encoding other than UTF-8, you can set spring.banner.charset. In addition to a text file, you can also add a banner.gif, banner.jpg, or banner.png image file to your classpath or set the spring.banner.image.location property. Images are converted into an ASCII art representation and printed above any text banner.

Inside your banner.txt file, you can use any of the following placeholders:

Table 1. Banner variables
Variable Description

${application.version}

The version number of your application, as declared in MANIFEST.MF. For example, Implementation-Version: 1.0 is printed as 1.0.

${application.formatted-version}

The version number of your application, as declared in MANIFEST.MF and formatted for display (surrounded with brackets and prefixed with v). For example (v1.0).

${spring-boot.version}

The Spring Boot version that you are using. For example 2.4.6.

${spring-boot.formatted-version}

The Spring Boot version that you are using, formatted for display (surrounded with brackets and prefixed with v). For example (v2.4.6).

${Ansi.NAME} (or ${AnsiColor.NAME}, ${AnsiBackground.NAME}, ${AnsiStyle.NAME})

Where NAME is the name of an ANSI escape code. See AnsiPropertySource for details.

${application.title}

The title of your application, as declared in MANIFEST.MF. For example Implementation-Title: MyApp is printed as MyApp.

The SpringApplication.setBanner(…) method can be used if you want to generate a banner programmatically. Use the org.springframework.boot.Banner interface and implement your own printBanner() method.

You can also use the spring.main.banner-mode property to determine if the banner has to be printed on System.out (console), sent to the configured logger (log), or not produced at all (off).

The printed banner is registered as a singleton bean under the following name: springBootBanner.

The ${application.version} and ${application.formatted-version} properties are only available if you are using Spring Boot launchers. The values won’t be resolved if you are running an unpacked jar and starting it with java -cp <classpath> <mainclass>.

This is why we recommend that you always launch unpacked jars using java org.springframework.boot.loader.JarLauncher. This will initialize the application.* banner variables before building the classpath and launching your app.

1.4. Customizing SpringApplication

If the SpringApplication defaults are not to your taste, you can instead create a local instance and customize it. For example, to turn off the banner, you could write:

public static void main(String[] args) {
    SpringApplication app = new SpringApplication(MySpringConfiguration.class);
    app.setBannerMode(Banner.Mode.OFF);
    app.run(args);
}
The constructor arguments passed to SpringApplication are configuration sources for Spring beans. In most cases, these are references to @Configuration classes, but they could also be references to XML configuration or to packages that should be scanned.

It is also possible to configure the SpringApplication by using an application.properties file. See Externalized Configuration for details.

For a complete list of the configuration options, see the SpringApplication Javadoc .

1.5. Fluent Builder API

If you need to build an ApplicationContext hierarchy (multiple contexts with a parent/child relationship) or if you prefer using a “fluent” builder API, you can use the SpringApplicationBuilder.

The SpringApplicationBuilder lets you chain together multiple method calls and includes parent and child methods that let you create a hierarchy, as shown in the following example:

new SpringApplicationBuilder()
        .sources(Parent.class)
        .child(Application.class)
        .bannerMode(Banner.Mode.OFF)
        .run(args);
There are some restrictions when creating an ApplicationContext hierarchy. For example, Web components must be contained within the child context, and the same Environment is used for both parent and child contexts. See the SpringApplicationBuilder Javadoc for full details.

1.6. Application Availability

When deployed on platforms, applications can provide information about their availability to the platform using infrastructure such as Kubernetes Probes . Spring Boot includes out-of-the box support for the commonly used “liveness” and “readiness” availability states. If you are using Spring Boot’s “actuator” support then these states are exposed as health endpoint groups.

In addition, you can also obtain availability states by injecting the ApplicationAvailability interface into your own beans.

1.6.1. Liveness State

The “Liveness” state of an application tells whether its internal state allows it to work correctly, or recover by itself if it’s currently failing. A broken “Liveness” state means that the application is in a state that it cannot recover from, and the infrastructure should restart the application.

In general, the "Liveness" state should not be based on external checks, such as Health checks. If it did, a failing external system (a database, a Web API, an external cache) would trigger massive restarts and cascading failures across the platform.

The internal state of Spring Boot applications is mostly represented by the Spring ApplicationContext. If the application context has started successfully, Spring Boot assumes that the application is in a valid state. An application is considered live as soon as the context has been refreshed, see Spring Boot application lifecycle and related Application Events.

1.6.2. Readiness State

The “Readiness” state of an application tells whether the application is ready to handle traffic. A failing “Readiness” state tells the platform that it should not route traffic to the application for now. This typically happens during startup, while CommandLineRunner and ApplicationRunner components are being processed, or at any time if the application decides that it’s too busy for additional traffic.

An application is considered ready as soon as application and command-line runners have been called, see Spring Boot application lifecycle and related Application Events.

Tasks expected to run during startup should be executed by CommandLineRunner and ApplicationRunner components instead of using Spring component lifecycle callbacks such as @PostConstruct.

1.6.3. Managing the Application Availability State

Application components can retrieve the current availability state at any time, by injecting the ApplicationAvailability interface and calling methods on it. More often, applications will want to listen to state updates or update the state of the application.

For example, we can export the "Readiness" state of the application to a file so that a Kubernetes "exec Probe" can look at this file:

@Component
public class ReadinessStateExporter {

    @EventListener
    public void onStateChange(AvailabilityChangeEvent<ReadinessState> event) {
        switch (event.getState()) {
        case ACCEPTING_TRAFFIC:
            // create file /tmp/healthy
        break;
        case REFUSING_TRAFFIC:
            // remove file /tmp/healthy
        break;
        }
    }

}

We can also update the state of the application, when the application breaks and cannot recover:

@Component
public class LocalCacheVerifier {

    private final ApplicationEventPublisher eventPublisher;

    public LocalCacheVerifier(ApplicationEventPublisher eventPublisher) {
        this.eventPublisher = eventPublisher;
    }

    public void checkLocalCache() {
        try {
            //...
        }
        catch (CacheCompletelyBrokenException ex) {
            AvailabilityChangeEvent.publish(this.eventPublisher, ex, LivenessState.BROKEN);
        }
    }

}

1.7. Application Events and Listeners

In addition to the usual Spring Framework events, such as ContextRefreshedEvent , a SpringApplication sends some additional application events.

Some events are actually triggered before the ApplicationContext is created, so you cannot register a listener on those as a @Bean. You can register them with the SpringApplication.addListeners(…) method or the SpringApplicationBuilder.listeners(…) method.

If you want those listeners to be registered automatically, regardless of the way the application is created, you can add a META-INF/spring.factories file to your project and reference your listener(s) by using the org.springframework.context.ApplicationListener key, as shown in the following example:

org.springframework.context.ApplicationListener=com.example.project.MyListener

Application events are sent in the following order, as your application runs:

  1. An ApplicationStartingEvent is sent at the start of a run but before any processing, except for the registration of listeners and initializers.

  2. An ApplicationEnvironmentPreparedEvent is sent when the Environment to be used in the context is known but before the context is created.

  3. An ApplicationContextInitializedEvent is sent when the ApplicationContext is prepared and ApplicationContextInitializers have been called but before any bean definitions are loaded.

  4. An ApplicationPreparedEvent is sent just before the refresh is started but after bean definitions have been loaded.

  5. An ApplicationStartedEvent is sent after the context has been refreshed but before any application and command-line runners have been called.

  6. An AvailabilityChangeEvent is sent right after with LivenessState.CORRECT to indicate that the application is considered as live.

  7. An ApplicationReadyEvent is sent after any application and command-line runners have been called.

  8. An AvailabilityChangeEvent is sent right after with ReadinessState.ACCEPTING_TRAFFIC to indicate that the application is ready to service requests.

  9. An ApplicationFailedEvent is sent if there is an exception on startup.

The above list only includes SpringApplicationEvents that are tied to a SpringApplication. In addition to these, the following events are also published after ApplicationPreparedEvent and before ApplicationStartedEvent:

  • A WebServerInitializedEvent is sent after the WebServer is ready. ServletWebServerInitializedEvent and ReactiveWebServerInitializedEvent are the servlet and reactive variants respectively.

  • A ContextRefreshedEvent is sent when an ApplicationContext is refreshed.

You often need not use application events, but it can be handy to know that they exist. Internally, Spring Boot uses events to handle a variety of tasks.
Event listeners should not run potentially lengthy tasks as they execute in the same thread by default. Consider using application and command-line runners instead.

Application events are sent by using Spring Framework’s event publishing mechanism. Part of this mechanism ensures that an event published to the listeners in a child context is also published to the listeners in any ancestor contexts. As a result of this, if your application uses a hierarchy of SpringApplication instances, a listener may receive multiple instances of the same type of application event.

To allow your listener to distinguish between an event for its context and an event for a descendant context, it should request that its application context is injected and then compare the injected context with the context of the event. The context can be injected by implementing ApplicationContextAware or, if the listener is a bean, by using @Autowired.

1.8. Web Environment

A SpringApplication attempts to create the right type of ApplicationContext on your behalf. The algorithm used to determine a WebApplicationType is the following:

  • If Spring MVC is present, an AnnotationConfigServletWebServerApplicationContext is used

  • If Spring MVC is not present and Spring WebFlux is present, an AnnotationConfigReactiveWebServerApplicationContext is used

  • Otherwise, AnnotationConfigApplicationContext is used

This means that if you are using Spring MVC and the new WebClient from Spring WebFlux in the same application, Spring MVC will be used by default. You can override that easily by calling setWebApplicationType(WebApplicationType).

It is also possible to take complete control of the ApplicationContext type that is used by calling setApplicationContextClass(…).

It is often desirable to call setWebApplicationType(WebApplicationType.NONE) when using SpringApplication within a JUnit test.

1.9. Accessing Application Arguments

If you need to access the application arguments that were passed to SpringApplication.run(…), you can inject a org.springframework.boot.ApplicationArguments bean. The ApplicationArguments interface provides access to both the raw String[] arguments as well as parsed option and non-option arguments, as shown in the following example:

import org.springframework.boot.*;
import org.springframework.beans.factory.annotation.*;
import org.springframework.stereotype.*;

@Component
public class MyBean {

    @Autowired
    public MyBean(ApplicationArguments args) {
        boolean debug = args.containsOption("debug");
        List<String> files = args.getNonOptionArgs();
        // if run with "--debug logfile.txt" debug=true, files=["logfile.txt"]
    }

}
Spring Boot also registers a CommandLinePropertySource with the Spring Environment. This lets you also inject single application arguments by using the @Value annotation.

1.10. Using the ApplicationRunner or CommandLineRunner

If you need to run some specific code once the SpringApplication has started, you can implement the ApplicationRunner or CommandLineRunner interfaces. Both interfaces work in the same way and offer a single run method, which is called just before SpringApplication.run(…) completes.

This contract is well suited for tasks that should run after application startup but before it starts accepting traffic.

The CommandLineRunner interfaces provides access to application arguments as a string array, whereas the ApplicationRunner uses the ApplicationArguments interface discussed earlier. The following example shows a CommandLineRunner with a run method:

import org.springframework.boot.*;
import org.springframework.stereotype.*;

@Component
public class MyBean implements CommandLineRunner {

    public void run(String... args) {
        // Do something...
    }

}

If several CommandLineRunner or ApplicationRunner beans are defined that must be called in a specific order, you can additionally implement the org.springframework.core.Ordered interface or use the org.springframework.core.annotation.Order annotation.

1.11. Application Exit

Each SpringApplication registers a shutdown hook with the JVM to ensure that the ApplicationContext closes gracefully on exit. All the standard Spring lifecycle callbacks (such as the DisposableBean interface or the @PreDestroy annotation) can be used.

In addition, beans may implement the org.springframework.boot.ExitCodeGenerator interface if they wish to return a specific exit code when SpringApplication.exit() is called. This exit code can then be passed to System.exit() to return it as a status code, as shown in the following example:

@SpringBootApplication
public class ExitCodeApplication {

    @Bean
    public ExitCodeGenerator exitCodeGenerator() {
        return () -> 42;
    }

    public static void main(String[] args) {
        System.exit(SpringApplication.exit(SpringApplication.run(ExitCodeApplication.class, args)));
    }

}

Also, the ExitCodeGenerator interface may be implemented by exceptions. When such an exception is encountered, Spring Boot returns the exit code provided by the implemented getExitCode() method.

1.12. Admin Features

It is possible to enable admin-related features for the application by specifying the spring.application.admin.enabled property. This exposes the SpringApplicationAdminMXBean on the platform MBeanServer. You could use this feature to administer your Spring Boot application remotely. This feature could also be useful for any service wrapper implementation.

If you want to know on which HTTP port the application is running, get the property with a key of local.server.port.

1.13. Application Startup tracking

During the application startup, the SpringApplication and the ApplicationContext perform many tasks related to the application lifecycle, the beans lifecycle or even processing application events. With ApplicationStartup , Spring Framework allows you to track the application startup sequence with StartupSteps . This data can be collected for profiling purposes, or just to have a better understanding of an application startup process.

You can choose an ApplicationStartup implementation when setting up the SpringApplication instance. For example, to use the BufferingApplicationStartup, you could write:

public static void main(String[] args) {
    SpringApplication app = new SpringApplication(MySpringConfiguration.class);
    app.setApplicationStartup(new BufferingApplicationStartup(2048));
    app.run(args);
}

The first available implementation, FlightRecorderApplicationStartup is provided by Spring Framework. It adds Spring-specific startup events to a Java Flight Recorder session and is meant for profiling applications and correlating their Spring context lifecycle with JVM events (such as allocations, GCs, class loading…). Once configured, you can record data by running the application with the Flight Recorder enabled:

$ java -XX:StartFlightRecording:filename=recording.jfr,duration=10s -jar demo.jar

Spring Boot ships with the BufferingApplicationStartup variant; this implementation is meant for buffering the startup steps and draining them into an external metrics system. Applications can ask for the bean of type BufferingApplicationStartup in any component. Additionally, Spring Boot Actuator will expose a startup endpoint to expose this information as a JSON document .

2. Externalized Configuration

Spring Boot lets you externalize your configuration so that you can work with the same application code in different environments. You can use a variety of external configuration sources, include Java properties files, YAML files, environment variables, and command-line arguments.

Property values can be injected directly into your beans by using the @Value annotation, accessed through Spring’s Environment abstraction, or be bound to structured objects through @ConfigurationProperties.

Spring Boot uses a very particular PropertySource order that is designed to allow sensible overriding of values. Properties are considered in the following order (with values from lower items overriding earlier ones):

  1. Default properties (specified by setting SpringApplication.setDefaultProperties).

  2. @PropertySource annotations on your @Configuration classes. Please note that such property sources are not added to the Environment until the application context is being refreshed. This is too late to configure certain properties such as logging.* and spring.main.* which are read before refresh begins.

  3. Config data (such as application.properties files)

  4. A RandomValuePropertySource that has properties only in random.*.

  5. OS environment variables.

  6. Java System properties (System.getProperties()).

  7. JNDI attributes from java:comp/env.

  8. ServletContext init parameters.

  9. ServletConfig init parameters.

  10. Properties from SPRING_APPLICATION_JSON (inline JSON embedded in an environment variable or system property).

  11. Command line arguments.

  12. properties attribute on your tests. Available on @SpringBootTest and the test annotations for testing a particular slice of your application.

  13. @TestPropertySource annotations on your tests.

  14. Devtools global settings properties in the $HOME/.config/spring-boot directory when devtools is active.

Config data files are considered in the following order:

  1. Application properties packaged inside your jar (application.properties and YAML variants).

  2. Profile-specific application properties packaged inside your jar (application-{profile}.properties and YAML variants).

  3. Application properties outside of your packaged jar (application.properties and YAML variants).

  4. Profile-specific application properties outside of your packaged jar (application-{profile}.properties and YAML variants).

It is recommended to stick with one format for your entire application. If you have configuration files with both .properties and .yml format in the same location, .properties takes precedence.

To provide a concrete example, suppose you develop a @Component that uses a name property, as shown in the following example:

import org.springframework.stereotype.*;
import org.springframework.beans.factory.annotation.*;

@Component
public class MyBean {

    @Value("${name}")
    private String name;

    // ...

}

On your application classpath (for example, inside your jar) you can have an application.properties file that provides a sensible default property value for name. When running in a new environment, an application.properties file can be provided outside of your jar that overrides the name. For one-off testing, you can launch with a specific command line switch (for example, java -jar app.jar --name="Spring").

The env and configprops endpoints can be useful in determining why a property has a particular value. You can use these two endpoints to diagnose unexpected property values. See the "Production ready features" section for details.

2.1. Accessing Command Line Properties

By default, SpringApplication converts any command line option arguments (that is, arguments starting with --, such as --server.port=9000) to a property and adds them to the Spring Environment. As mentioned previously, command line properties always take precedence over file based property sources.

If you do not want command line properties to be added to the Environment, you can disable them by using SpringApplication.setAddCommandLineProperties(false).

2.2. JSON Application Properties

Environment variables and system properties often have restrictions that mean some property names cannot be used. To help with this, Spring Boot allows you to encode a block of properties into a single JSON structure.

When your application starts, any spring.application.json or SPRING_APPLICATION_JSON properties will be parsed and added to the Environment.

For example, the SPRING_APPLICATION_JSON property can be supplied on the command line in a UN*X shell as an environment variable:

$ SPRING_APPLICATION_JSON='{"acme":{"name":"test"}}' java -jar myapp.jar

In the preceding example, you end up with acme.name=test in the Spring Environment.

The same JSON can also be provided as a system property:

$ java -Dspring.application.json='{"acme":{"name":"test"}}' -jar myapp.jar

Or you could supply the JSON by using a command line argument:

$ java -jar myapp.jar --spring.application.json='{"acme":{"name":"test"}}'

If you are deploying to a classic Application Server, you could also use a JNDI variable named java:comp/env/spring.application.json.

Although null values from the JSON will be added to the resulting property source, the PropertySourcesPropertyResolver treats null properties as missing values. This means that the JSON cannot override properties from lower order property sources with a null value.

2.3. External Application Properties

Spring Boot will automatically find and load application.properties and application.yaml files from the following locations when your application starts:

  1. The classpath root

  2. The classpath /config package

  3. The current directory

  4. The /config subdirectory in the current directory

  5. Immediate child directories of the /config subdirectory

The list is ordered by precedence (with values from lower items overriding earlier ones). Documents from the loaded files are added as PropertySources to the Spring Environment.

If you do not like application as the configuration file name, you can switch to another file name by specifying a spring.config.name environment property. You can also refer to an explicit location by using the spring.config.location environment property (which is a comma-separated list of directory locations or file paths). The following example shows how to specify a different file name:

$ java -jar myproject.jar --spring.config.name=myproject

The following example shows how to specify two locations:

$ java -jar myproject.jar --spring.config.location=optional:classpath:/default.properties,optional:classpath:/override.properties
Use the prefix optional: if the locations are optional and you don’t mind if they don’t exist.
spring.config.name, spring.config.location, and spring.config.additional-location are used very early to determine which files have to be loaded. They must be defined as an environment property (typically an OS environment variable, a system property, or a command-line argument).

If spring.config.location contains directories (as opposed to files), they should end in / (at runtime they will be appended with the names generated from spring.config.name before being loaded). Files specified in spring.config.location are used as-is. Whether specified directly or contained in a directory, configuration files must include a file extension in their name. Typical extensions that are supported out-of-the-box are .properties, .yaml, and .yml.

When multiple locations are specified, the later ones can override the values of earlier ones.

Locations configured by using spring.config.location replace the default locations. For example, if spring.config.location is configured with the value optional:classpath:/custom-config/,optional:file:./custom-config/, the complete set of locations considered is:

  1. optional:classpath:custom-config/

  2. optional:file:./custom-config/

If you prefer to add additional locations, rather than replacing them, you can use spring.config.additional-location. Properties loaded from additional locations can override those in the default locations. For example, if spring.config.additional-location is configured with the value optional:classpath:/custom-config/,optional:file:./custom-config/, the complete set of locations considered is:

  1. optional:classpath:/

  2. optional:classpath:/config/

  3. optional:file:./

  4. optional:file:./config/

  5. optional:file:./config/*/

  6. optional:classpath:custom-config/

  7. optional:file:./custom-config/

This search ordering lets you specify default values in one configuration file and then selectively override those values in another. You can provide default values for your application in application.properties (or whatever other basename you choose with spring.config.name) in one of the default locations. These default values can then be overridden at runtime with a different file located in one of the custom locations.

If you use environment variables rather than system properties, most operating systems disallow period-separated key names, but you can use underscores instead (for example, SPRING_CONFIG_NAME instead of spring.config.name). See Binding from Environment Variables for details.
If your application runs in a servlet container or application server, then JNDI properties (in java:comp/env) or servlet context initialization parameters can be used instead of, or as well as, environment variables or system properties.

2.3.1. Optional Locations

By default, when a specified config data location does not exist, Spring Boot will throw a ConfigDataLocationNotFoundException and your application will not start.

If you want to specify a location, but you don’t mind if it doesn’t always exist, you can use the optional: prefix. You can use this prefix with the spring.config.location and spring.config.additional-location properties, as well as with spring.config.import declarations.

For example, a spring.config.import value of optional:file:./myconfig.properties allows your application to start, even if the myconfig.properties file is missing.

If you want to ignore all ConfigDataLocationNotFoundExceptions and always continue to start your application, you can use the spring.config.on-not-found property. Set the value to ignore using SpringApplication.setDefaultProperties(…) or with a system/environment variable.

2.3.2. Wildcard Locations

If a config file location includes the * character for the last path segment, it is considered a wildcard location. Wildcards are expanded when the config is loaded so that immediate subdirectories are also checked. Wildcard locations are particularly useful in an environment such as Kubernetes when there are multiple sources of config properties.

For example, if you have some Redis configuration and some MySQL configuration, you might want to keep those two pieces of configuration separate, while requiring that both those are present in an application.properties file. This might result in two separate application.properties files mounted at different locations such as /config/redis/application.properties and /config/mysql/application.properties. In such a case, having a wildcard location of config/*/, will result in both files being processed.

By default, Spring Boot includes config/*/ in the default search locations. It means that all subdirectories of the /config directory outside of your jar will be searched.

You can use wildcard locations yourself with the spring.config.location and spring.config.additional-location properties.

A wildcard location must contain only one * and end with */ for search locations that are directories or */<filename> for search locations that are files. Locations with wildcards are sorted alphabetically based on the absolute path of the file names.
Wildcard locations only work with external directories. You cannot use a wildcard in a classpath: location.

2.3.3. Profile Specific Files

As well as application property files, Spring Boot will also attempt to load profile-specific files using the naming convention application-{profile}. For example, if your application activates a profile named prod and uses YAML files, then both application.yml and application-prod.yml will be considered.

Profile-specific properties are loaded from the same locations as standard application.properties, with profile-specific files always overriding the non-specific ones. If several profiles are specified, a last-wins strategy applies. For example, if profiles prod,live are specified by the spring.profiles.active property, values in application-prod.properties can be overridden by those in application-live.properties.

The Environment has a set of default profiles (by default, [default]) that are used if no active profiles are set. In other words, if no profiles are explicitly activated, then properties from application-default are considered.

Properties files are only ever loaded once. If you’ve already directly imported a profile specific property files then it won’t be imported a second time.

2.3.4. Importing Additional Data

Application properties may import further config data from other locations using the spring.config.import property. Imports are processed as they are discovered, and are treated as additional documents inserted immediately below the one that declares the import.

For example, you might have the following in your classpath application.properties file:

Properties
spring.application.name=myapp
spring.config.import=optional:file:./dev.properties
Yaml
spring:
  application:
    name: "myapp"
  config:
    import: "optional:file:./dev.properties"

This will trigger the import of a dev.properties file in current directory (if such a file exists). Values from the imported dev.properties will take precedence over the file that triggered the import. In the above example, the dev.properties could redefine spring.application.name to a different value. An import will only be imported once no matter how many times it is declared. The order an import is defined inside a single document within the properties/yaml file doesn’t matter. For instance, the two examples below produce the same result:

Properties
spring.config.import=my.properties
my.property=value
Yaml
spring:
  config:
    import: my.properties
my:
  property: value
Properties
my.property=value
spring.config.import=my.properties
Yaml
my:
  property: value
spring:
  config:
    import: my.properties

In both of the above examples, the values from the my.properties file will take precedence over the file that triggered its import.

Several locations can be specified under a single spring.config.import key. Locations will be processed in the order that they are defined, with later imports taking precedence.

Spring Boot includes pluggable API that allows various different location addresses to be supported. By default you can import Java Properties, YAML and “configuration trees”.

Third-party jars can offer support for additional technologies (there’s no requirement for files to be local). For example, you can imagine config data being from external stores such as Consul, Apache ZooKeeper or Netflix Archaius.

If you want to support your own locations, see the ConfigDataLocationResolver and ConfigDataLoader classes in the org.springframework.boot.context.config package.

2.3.5. Importing Extensionless Files

Some cloud platforms cannot add a file extension to volume mounted files. To import these extensionless files, you need to give Spring Boot a hint so that it knows how to load them. You can do this by putting an extension hint in square brackets.

For example, suppose you have a /etc/config/myconfig file that you wish to import as yaml. You can import it from your application.properties using the following:

Properties
spring.config.import=file:/etc/config/myconfig[.yaml]
Yaml
spring:
  config:
    import: "file:/etc/config/myconfig[.yaml]"

2.3.6. Using Configuration Trees

When running applications on a cloud platform (such as Kubernetes) you often need to read config values that the platform supplies. It’s not uncommon to use environment variables for such purposes, but this can have drawbacks, especially if the value is supposed to be kept secret.

As an alternative to environment variables, many cloud platforms now allow you to map configuration into mounted data volumes. For example, Kubernetes can volume mount both ConfigMaps and Secrets .

There are two common volume mount patterns that can be use:

  1. A single file contains a complete set of properties (usually written as YAML).

  2. Multiple files are written to a directory tree, with the filename becoming the ‘key’ and the contents becoming the ‘value’.

For the first case, you can import the YAML or Properties file directly using spring.config.import as described above. For the second case, you need to use the configtree: prefix so that Spring Boot knows it needs to expose all the files as properties.

As an example, let’s imagine that Kubernetes has mounted the following volume:

etc/
  config/
    myapp/
      username
      password

The contents of the username file would be a config value, and the contents of password would be a secret.

To import these properties, you can add the following to your application.properties or application.yaml file:

Properties
spring.config.import=optional:configtree:/etc/config/
Yaml
spring:
  config:
    import: "optional:configtree:/etc/config/"

You can then access or inject myapp.username and myapp.password properties from the Environment in the usual way.

Configuration tree values can be bound to both string String and byte[] types depending on the contents expected.

If you have multiple config trees to import from the same parent folder you can use a wildcard shortcut. Any configtree: location that ends with /*/ will import all immediate children as config trees.

For example, given the following volume:

etc/
  config/
    dbconfig/
      db/
        username
        password
    mqconfig/
      mq/
        username
        password

You can use configtree:/etc/config/*/ as the import location:

Properties
spring.config.import=optional:configtree:/etc/config/*/
Yaml
spring:
  config:
    import: "optional:configtree:/etc/config/*/"

This will add db.username, db.password, mq.username and mq.password properties.

Directories loaded using a wildcard are sorted alphabetically. If you need a different order, then you should list each location as a separate import

Configuration trees can also be used for Docker secrets. When a Docker swarm service is granted access to a secret, the secret gets mounted into the container. For example, if a secret named db.password is mounted at location /run/secrets/, you can make db.password available to the Spring environment using the following:

Properties
spring.config.import=optional:configtree:/run/secrets/
Yaml
spring:
  config:
    import: "optional:configtree:/run/secrets/"

2.3.7. Property Placeholders

The values in application.properties and application.yml are filtered through the existing Environment when they are used, so you can refer back to previously defined values (for example, from System properties). The standard ${name} property-placeholder syntax can be used anywhere within a value.

For example, the following file will set app.description to “MyApp is a Spring Boot application”:

Properties
app.name=MyApp
app.description=${app.name} is a Spring Boot application
Yaml
app:
  name: "MyApp"
  description: "${app.name} is a Spring Boot application"
You can also use this technique to create “short” variants of existing Spring Boot properties. See the howto.html how-to for details.

2.3.8. Working with Multi-Document Files

Spring Boot allows you to split a single physical file into multiple logical documents which are each added independently. Documents are processed in order, from top to bottom. Later documents can override the properties defined in earlier ones.

For application.yml files, the standard YAML multi-document syntax is used. Three consecutive hyphens represent the end of one document, and the start of the next.

For example, the following file has two logical documents:

spring.application.name: MyApp
---
spring.config.activate.on-cloud-platform: kubernetes
spring.application.name: MyCloudApp

For application.properties files a special #--- comment is used to mark the document splits:

spring.application.name=MyApp
#---
spring.config.activate.on-cloud-platform=kubernetes
spring.application.name=MyCloudApp
Property file separators must not have any leading whitespace and must have exactly three hyphen characters. The lines immediately before and after the separator must not be comments.
Multi-document property files are often used in conjunction with activation properties such as spring.config.activate.on-profile. See the next section for details.
Multi-document property files cannot be loaded by using the @PropertySource or @TestPropertySource annotations.

2.3.9. Activation Properties

It’s sometimes useful to only activate a given get of properties when certain conditions are met. For example, you might have properties that are only relevant when a specific profile is active.

You can conditionally activate a properties document using spring.config.activate.*.

The following activation properties are available:

Table 2. activation properties
Property Note

on-profile

A profile expression that must match for the document to be active.

on-cloud-platform

The CloudPlatform that must be detected for the document to be active.

For example, the following specifies that the second document is only active when running on Kubernetes, and only when either the “prod” or “staging” profiles are active:

Properties
myprop=always-set
#---
spring.config.activate.on-cloud-platform=kubernetes
spring.config.activate.on-profile=prod | staging
myotherprop=sometimes-set
Yaml
myprop:
  always-set
---
spring:
  config:
    activate:
      on-cloud-platform: "kubernetes"
      on-profile: "prod | staging"
myotherprop: sometimes-set

2.4. Encrypting Properties

Spring Boot does not provide any built in support for encrypting property values, however, it does provide the hook points necessary to modify values contained in the Spring Environment. The EnvironmentPostProcessor interface allows you to manipulate the Environment before the application starts. See howto.html for details.

If you’re looking for a secure way to store credentials and passwords, the Spring Cloud Vault project provides support for storing externalized configuration in HashiCorp Vault .

2.5. Working with YAML

YAML is a superset of JSON and, as such, is a convenient format for specifying hierarchical configuration data. The SpringApplication class automatically supports YAML as an alternative to properties whenever you have the SnakeYAML library on your classpath.

If you use “Starters”, SnakeYAML is automatically provided by spring-boot-starter.

2.5.1. Mapping YAML to Properties

YAML documents need to be converted from their hierarchical format to a flat structure that can be used with the Spring Environment. For example, consider the following YAML document:

environments:
  dev:
    url: https://dev.example.com
    name: Developer Setup
  prod:
    url: https://another.example.com
    name: My Cool App

In order to access these properties from the Environment, they would be flattened as follows:

environments.dev.url=https://dev.example.com
environments.dev.name=Developer Setup
environments.prod.url=https://another.example.com
environments.prod.name=My Cool App

Likewise, YAML lists also need to be flattened. They are represented as property keys with [index] dereferencers. For example, consider the following YAML:

my:
 servers:
 - dev.example.com
 - another.example.com

The preceding example would be transformed into these properties:

my.servers[0]=dev.example.com
my.servers[1]=another.example.com
Properties that use the [index] notation can be bound to Java List or Set objects using Spring Boot’s Binder class. For more details see the “Type-safe Configuration Properties” section below.
YAML files cannot be loaded by using the @PropertySource or @TestPropertySource annotations. So, in the case that you need to load values that way, you need to use a properties file.

2.5.2. Directly Loading YAML

Spring Framework provides two convenient classes that can be used to load YAML documents. The YamlPropertiesFactoryBean loads YAML as Properties and the YamlMapFactoryBean loads YAML as a Map.

You can also use the YamlPropertySourceLoader class if you want to load YAML as a Spring PropertySource.

2.6. Configuring Random Values

The RandomValuePropertySource is useful for injecting random values (for example, into secrets or test cases). It can produce integers, longs, uuids, or strings, as shown in the following example:

Properties
my.secret=${random.value}
my.number=${random.int}
my.bignumber=${random.long}
my.uuid=${random.uuid}
my.number-less-than-ten=${random.int(10)}
my.number-in-range=${random.int[1024,65536]}
Yaml
my:
  secret: "${random.value}"
  number: "${random.int}"
  bignumber: "${random.long}"
  uuid: "${random.uuid}"
  number-less-than-ten: "${random.int(10)}"
  number-in-range: "${random.int[1024,65536]}"

The random.int* syntax is OPEN value (,max) CLOSE where the OPEN,CLOSE are any character and value,max are integers. If max is provided, then value is the minimum value and max is the maximum value (exclusive).

2.7. Type-safe Configuration Properties

Using the @Value("${property}") annotation to inject configuration properties can sometimes be cumbersome, especially if you are working with multiple properties or your data is hierarchical in nature. Spring Boot provides an alternative method of working with properties that lets strongly typed beans govern and validate the configuration of your application.

2.7.1. JavaBean properties binding

It is possible to bind a bean declaring standard JavaBean properties as shown in the following example:

package com.example;

import java.net.InetAddress;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;

import org.springframework.boot.context.properties.ConfigurationProperties;

@ConfigurationProperties("acme")
public class AcmeProperties {

    private boolean enabled;

    private InetAddress remoteAddress;

    private final Security security = new Security();

    public boolean isEnabled() { ... }

    public void setEnabled(boolean enabled) { ... }

    public InetAddress getRemoteAddress() { ... }

    public void setRemoteAddress(InetAddress remoteAddress) { ... }

    public Security getSecurity() { ... }

    public static class Security {

        private String username;

        private String password;

        private List<String> roles = new ArrayList<>(Collections.singleton("USER"));

        public String getUsername() { ... }

        public void setUsername(String username) { ... }

        public String getPassword() { ... }

        public void setPassword(String password) { ... }

        public List<String> getRoles() { ... }

        public void setRoles(List<String> roles) { ... }

    }
}

The preceding POJO defines the following properties:

  • acme.enabled, with a value of false by default.

  • acme.remote-address, with a type that can be coerced from String.

  • acme.security.username, with a nested "security" object whose name is determined by the name of the property. In particular, the return type is not used at all there and could have been SecurityProperties.

  • acme.security.password.

  • acme.security.roles, with a collection of String that defaults to USER.

The properties that map to @ConfigurationProperties classes available in Spring Boot, which are configured via properties files, YAML files, environment variables etc., are public API but the accessors (getters/setters) of the class itself are not meant to be used directly.

Such arrangement relies on a default empty constructor and getters and setters are usually mandatory, since binding is through standard Java Beans property descriptors, just like in Spring MVC. A setter may be omitted in the following cases:

  • Maps, as long as they are initialized, need a getter but not necessarily a setter, since they can be mutated by the binder.

  • Collections and arrays can be accessed either through an index (typically with YAML) or by using a single comma-separated value (properties). In the latter case, a setter is mandatory. We recommend to always add a setter for such types. If you initialize a collection, make sure it is not immutable (as in the preceding example).

  • If nested POJO properties are initialized (like the Security field in the preceding example), a setter is not required. If you want the binder to create the instance on the fly by using its default constructor, you need a setter.

Some people use Project Lombok to add getters and setters automatically. Make sure that Lombok does not generate any particular constructor for such a type, as it is used automatically by the container to instantiate the object.

Finally, only standard Java Bean properties are considered and binding on static properties is not supported.

2.7.2. Constructor binding

The example in the previous section can be rewritten in an immutable fashion as shown in the following example:

package com.example;

import java.net.InetAddress;
import java.util.List;

import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.boot.context.properties.ConstructorBinding;
import org.springframework.boot.context.properties.bind.DefaultValue;

@ConstructorBinding
@ConfigurationProperties("acme")
public class AcmeProperties {

    private final boolean enabled;

    private final InetAddress remoteAddress;

    private final Security security;

    public AcmeProperties(boolean enabled, InetAddress remoteAddress, Security security) {
        this.enabled = enabled;
        this.remoteAddress = remoteAddress;
        this.security = security;
    }

    public boolean isEnabled() { ... }

    public InetAddress getRemoteAddress() { ... }

    public Security getSecurity() { ... }

    public static class Security {

        private final String username;

        private final String password;

        private final List<String> roles;

        public Security(String username, String password,
                @DefaultValue("USER") List<String> roles) {
            this.username = username;
            this.password = password;
            this.roles = roles;
        }

        public String getUsername() { ... }

        public String getPassword() { ... }

        public List<String> getRoles() { ... }

    }

}

In this setup, the @ConstructorBinding annotation is used to indicate that constructor binding should be used. This means that the binder will expect to find a constructor with the parameters that you wish to have bound.

Nested members of a @ConstructorBinding class (such as Security in the example above) will also be bound via their constructor.

Default values can be specified using @DefaultValue and the same conversion service will be applied to coerce the String value to the target type of a missing property. By default, if no properties are bound to Security, the AcmeProperties instance will contain a null value for security. If you wish you return a non-null instance of Security even when no properties are bound to it, you can use an empty @DefaultValue annotation to do so:

package com.example;
import java.net.InetAddress;
import java.util.List;

import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.boot.context.properties.ConstructorBinding;
import org.springframework.boot.context.properties.bind.DefaultValue;

@ConstructorBinding
@ConfigurationProperties("acme")
public class AcmeProperties {

    private final boolean enabled;

    private final InetAddress remoteAddress;

    private final Security security;

    public AcmeProperties(boolean enabled, InetAddress remoteAddress, @DefaultValue Security security) {
        this.enabled = enabled;
        this.remoteAddress = remoteAddress;
        this.security = security;
    }
}
To use constructor binding the class must be enabled using @EnableConfigurationProperties or configuration property scanning. You cannot use constructor binding with beans that are created by the regular Spring mechanisms (e.g. @Component beans, beans created via @Bean methods or beans loaded using @Import)
If you have more than one constructor for your class you can also use @ConstructorBinding directly on the constructor that should be bound.
The use of java.util.Optional with @ConfigurationProperties is not recommended as it is primarily intended for use as a return type. As such, it is not well-suited to configuration property injection. For consistency with properties of other types, if you do declare an Optional property and it has no value, null rather than an empty Optional will be bound.

2.7.3. Enabling @ConfigurationProperties-annotated types

Spring Boot provides infrastructure to bind @ConfigurationProperties types and register them as beans. You can either enable configuration properties on a class-by-class basis or enable configuration property scanning that works in a similar manner to component scanning.

Sometimes, classes annotated with @ConfigurationProperties might not be suitable for scanning, for example, if you’re developing your own auto-configuration or you want to enable them conditionally. In these cases, specify the list of types to process using the @EnableConfigurationProperties annotation. This can be done on any @Configuration class, as shown in the following example:

@Configuration(proxyBeanMethods = false)
@EnableConfigurationProperties(AcmeProperties.class)
public class MyConfiguration {
}

To use configuration property scanning, add the @ConfigurationPropertiesScan annotation to your application. Typically, it is added to the main application class that is annotated with @SpringBootApplication but it can be added to any @Configuration class. By default, scanning will occur from the package of the class that declares the annotation. If you want to define specific packages to scan, you can do so as shown in the following example:

@SpringBootApplication
@ConfigurationPropertiesScan({ "com.example.app", "org.acme.another" })
public class MyApplication {
}

When the @ConfigurationProperties bean is registered using configuration property scanning or via @EnableConfigurationProperties, the bean has a conventional name: <prefix>-<fqn>, where <prefix> is the environment key prefix specified in the @ConfigurationProperties annotation and <fqn> is the fully qualified name of the bean. If the annotation does not provide any prefix, only the fully qualified name of the bean is used.

The bean name in the example above is acme-com.example.AcmeProperties.

We recommend that @ConfigurationProperties only deal with the environment and, in particular, does not inject other beans from the context. For corner cases, setter injection can be used or any of the *Aware interfaces provided by the framework (such as EnvironmentAware if you need access to the Environment). If you still want to inject other beans using the constructor, the configuration properties bean must be annotated with @Component and use JavaBean-based property binding.

2.7.4. Using @ConfigurationProperties-annotated types

This style of configuration works particularly well with the SpringApplication external YAML configuration, as shown in the following example:

acme:
    remote-address: 192.168.1.1
    security:
        username: admin
        roles:
          - USER
          - ADMIN

To work with @ConfigurationProperties beans, you can inject them in the same way as any other bean, as shown in the following example:

@Service
public class MyService {

    private final AcmeProperties properties;

    @Autowired
    public MyService(AcmeProperties properties) {
        this.properties = properties;
    }

    //...

    @PostConstruct
    public void openConnection() {
        Server server = new Server(this.properties.getRemoteAddress());
        // ...
    }

}
Using @ConfigurationProperties also lets you generate metadata files that can be used by IDEs to offer auto-completion for your own keys. See the appendix for details.

2.7.5. Third-party Configuration

As well as using @ConfigurationProperties to annotate a class, you can also use it on public @Bean methods. Doing so can be particularly useful when you want to bind properties to third-party components that are outside of your control.

To configure a bean from the Environment properties, add @ConfigurationProperties to its bean registration, as shown in the following example:

@ConfigurationProperties(prefix = "another")
@Bean
public AnotherComponent anotherComponent() {
    ...
}

Any JavaBean property defined with the another prefix is mapped onto that AnotherComponent bean in manner similar to the preceding AcmeProperties example.

2.7.6. Relaxed Binding

Spring Boot uses some relaxed rules for binding Environment properties to @ConfigurationProperties beans, so there does not need to be an exact match between the Environment property name and the bean property name. Common examples where this is useful include dash-separated environment properties (for example, context-path binds to contextPath), and capitalized environment properties (for example, PORT binds to port).

As an example, consider the following @ConfigurationProperties class:

@ConfigurationProperties(prefix="acme.my-project.person")
public class OwnerProperties {

    private String firstName;

    public String getFirstName() {
        return this.firstName;
    }

    public void setFirstName(String firstName) {
        this.firstName = firstName;
    }

}

With the preceding code, the following properties names can all be used:

Table 3. relaxed binding
Property Note

acme.my-project.person.first-name

Kebab case, which is recommended for use in .properties and .yml files.

acme.myProject.person.firstName

Standard camel case syntax.

acme.my_project.person.first_name

Underscore notation, which is an alternative format for use in .properties and .yml files.

ACME_MYPROJECT_PERSON_FIRSTNAME

Upper case format, which is recommended when using system environment variables.

The prefix value for the annotation must be in kebab case (lowercase and separated by -, such as acme.my-project.person).
Table 4. relaxed binding rules per property source
Property Source Simple List

Properties Files

Camel case, kebab case, or underscore notation

Standard list syntax using [ ] or comma-separated values

YAML Files

Camel case, kebab case, or underscore notation

Standard YAML list syntax or comma-separated values

Environment Variables

Upper case format with underscore as the delimiter (see Binding from Environment Variables).

Numeric values surrounded by underscores (see Binding from Environment Variables)

System properties

Camel case, kebab case, or underscore notation

Standard list syntax using [ ] or comma-separated values

We recommend that, when possible, properties are stored in lower-case kebab format, such as my.property-name=acme.
Binding Maps

When binding to Map properties you may need to use a special bracket notation so that the original key value is preserved. If the key is not surrounded by [], any characters that are not alpha-numeric, - or . are removed.

For example, consider binding the following properties to a Map<String,String>:

Properties
acme.map.[/key1]=value1
acme.map.[/key2]=value2
acme.map./key3=value3
Yaml
acme:
  map:
    "[/key1]": "value1"
    "[/key2]": "value2"
    "/key3": "value3"
For YAML files, the brackets need to be surrounded by quotes for the keys to be parsed properly.

The properties above will bind to a Map with /key1, /key2 and key3 as the keys in the map. The slash has been removed from key3 because it wasn’t surrounded by square brackets.

You may also occasionally need to use the bracket notation if your key contains a . and you are binding to non-scalar value. For example, binding a.b=c to Map<String, Object> will return a Map with the entry {"a"={"b"="c"}} whereas [a.b]=c will return a Map with the entry {"a.b"="c"}.

Binding from Environment Variables

Most operating systems impose strict rules around the names that can be used for environment variables. For example, Linux shell variables can contain only letters (a to z or A to Z), numbers (0 to 9) or the underscore character (_). By convention, Unix shell variables will also have their names in UPPERCASE.

Spring Boot’s relaxed binding rules are, as much as possible, designed to be compatible with these naming restrictions.

To convert a property name in the canonical-form to an environment variable name you can follow these rules:

  • Replace dots (.) with underscores (_).

  • Remove any dashes (-).

  • Convert to uppercase.

For example, the configuration property spring.main.log-startup-info would be an environment variable named SPRING_MAIN_LOGSTARTUPINFO.

Environment variables can also be used when binding to object lists. To bind to a List, the element number should be surrounded with underscores in the variable name.

For example, the configuration property my.acme[0].other would use an environment variable named MY_ACME_0_OTHER.

2.7.7. Merging Complex Types

When lists are configured in more than one place, overriding works by replacing the entire list.

For example, assume a MyPojo object with name and description attributes that are null by default. The following example exposes a list of MyPojo objects from AcmeProperties:

@ConfigurationProperties("acme")
public class AcmeProperties {

    private final List<MyPojo> list = new ArrayList<>();

    public List<MyPojo> getList() {
        return this.list;
    }

}

Consider the following configuration:

Properties
acme.list[0].name=my name
acme.list[0].description=my description
#---
spring.config.activate.on-profile=dev
acme.list[0].name=my another name
Yaml
acme:
  list:
  - name: "my name"
    description: "my description"
---
spring:
  config:
    activate:
      on-profile: "dev"
acme:
  list:
  - name: "my another name"

If the dev profile is not active, AcmeProperties.list contains one MyPojo entry, as previously defined. If the dev profile is enabled, however, the list still contains only one entry (with a name of my another name and a description of null). This configuration does not add a second MyPojo instance to the list, and it does not merge the items.

When a List is specified in multiple profiles, the one with the highest priority (and only that one) is used. Consider the following example:

Properties
acme.list[0].name=my name
acme.list[0].description=my description
acme.list[1].name=another name
acme.list[1].description=another description
#---
spring.config.activate.on-profile=dev
acme.list[0].name=my another name
Yaml
acme:
  list:
  - name: "my name"
    description: "my description"
  - name: "another name"
    description: "another description"
---
spring:
  config:
    activate:
      on-profile: "dev"
acme:
  list:
  - name: "my another name"

In the preceding example, if the dev profile is active, AcmeProperties.list contains one MyPojo entry (with a name of my another name and a description of null). For YAML, both comma-separated lists and YAML lists can be used for completely overriding the contents of the list.

For Map properties, you can bind with property values drawn from multiple sources. However, for the same property in multiple sources, the one with the highest priority is used. The following example exposes a Map<String, MyPojo> from AcmeProperties:

@ConfigurationProperties("acme")
public class AcmeProperties {

    private final Map<String, MyPojo> map = new HashMap<>();

    public Map<String, MyPojo> getMap() {
        return this.map;
    }

}

Consider the following configuration:

Properties
acme.map.key1.name=my name 1
acme.map.key1.description=my description 1
#---
spring.config.activate.on-profile=dev
acme.map.key1.name=dev name 1
acme.map.key2.name=dev name 2
acme.map.key2.description=dev description 2
Yaml
acme:
  map:
    key1:
      name: "my name 1"
      description: "my description 1"
---
spring:
  config:
    activate:
      on-profile: "dev"
acme:
  map:
    key1:
      name: "dev name 1"
    key2:
      name: "dev name 2"
      description: "dev description 2"

If the dev profile is not active, AcmeProperties.map contains one entry with key key1 (with a name of my name 1 and a description of my description 1). If the dev profile is enabled, however, map contains two entries with keys key1 (with a name of dev name 1 and a description of my description 1) and key2 (with a name of dev name 2 and a description of dev description 2).

The preceding merging rules apply to properties from all property sources, and not just files.

2.7.8. Properties Conversion

Spring Boot attempts to coerce the external application properties to the right type when it binds to the @ConfigurationProperties beans. If you need custom type conversion, you can provide a ConversionService bean (with a bean named conversionService) or custom property editors (through a CustomEditorConfigurer bean) or custom Converters (with bean definitions annotated as @ConfigurationPropertiesBinding).

As this bean is requested very early during the application lifecycle, make sure to limit the dependencies that your ConversionService is using. Typically, any dependency that you require may not be fully initialized at creation time. You may want to rename your custom ConversionService if it is not required for configuration keys coercion and only rely on custom converters qualified with @ConfigurationPropertiesBinding.
Converting durations

Spring Boot has dedicated support for expressing durations. If you expose a java.time.Duration property, the following formats in application properties are available:

  • A regular long representation (using milliseconds as the default unit unless a @DurationUnit has been specified)

  • The standard ISO-8601 format used by java.time.Duration

  • A more readable format where the value and the unit are coupled (e.g. 10s means 10 seconds)

Consider the following example:

@ConfigurationProperties("app.system")
public class AppSystemProperties {

    @DurationUnit(ChronoUnit.SECONDS)
    private Duration sessionTimeout = Duration.ofSeconds(30);

    private Duration readTimeout = Duration.ofMillis(1000);

    public Duration getSessionTimeout() {
        return this.sessionTimeout;
    }

    public void setSessionTimeout(Duration sessionTimeout) {
        this.sessionTimeout = sessionTimeout;
    }

    public Duration getReadTimeout() {
        return this.readTimeout;
    }

    public void setReadTimeout(Duration readTimeout) {
        this.readTimeout = readTimeout;
    }

}

To specify a session timeout of 30 seconds, 30, PT30S and 30s are all equivalent. A read timeout of 500ms can be specified in any of the following form: 500, PT0.5S and 500ms.

You can also use any of the supported units. These are:

  • ns for nanoseconds

  • us for microseconds

  • ms for milliseconds

  • s for seconds

  • m for minutes

  • h for hours

  • d for days

The default unit is milliseconds and can be overridden using @DurationUnit as illustrated in the sample above.

If you prefer to use constructor binding, the same properties can be exposed, as shown in the following example:

@ConfigurationProperties("app.system")
@ConstructorBinding
public class AppSystemProperties {

    private final Duration sessionTimeout;

    private final Duration readTimeout;

    public AppSystemProperties(@DurationUnit(ChronoUnit.SECONDS) @DefaultValue("30s") Duration sessionTimeout,
            @DefaultValue("1000ms") Duration readTimeout) {
        this.sessionTimeout = sessionTimeout;
        this.readTimeout = readTimeout;
    }

    public Duration getSessionTimeout() {
        return this.sessionTimeout;
    }

    public Duration getReadTimeout() {
        return this.readTimeout;
    }

}
If you are upgrading a Long property, make sure to define the unit (using @DurationUnit) if it isn’t milliseconds. Doing so gives a transparent upgrade path while supporting a much richer format.
Converting periods

In addition to durations, Spring Boot can also work with java.time.Period type. The following formats can be used in application properties:

  • An regular int representation (using days as the default unit unless a @PeriodUnit has been specified)

  • The standard ISO-8601 format used by java.time.Period

  • A simpler format where the value and the unit pairs are coupled (e.g. 1y3d means 1 year and 3 days)

The following units are supported with the simple format:

  • y for years

  • m for months

  • w for weeks

  • d for days

The java.time.Period type never actually stores the number of weeks, it is a shortcut that means “7 days”.
Converting Data Sizes

Spring Framework has a DataSize value type that expresses a size in bytes. If you expose a DataSize property, the following formats in application properties are available:

  • A regular long representation (using bytes as the default unit unless a @DataSizeUnit has been specified)

  • A more readable format where the value and the unit are coupled (e.g. 10MB means 10 megabytes)

Consider the following example:

@ConfigurationProperties("app.io")
public class AppIoProperties {

    @DataSizeUnit(DataUnit.MEGABYTES)
    private DataSize bufferSize = DataSize.ofMegabytes(2);

    private DataSize sizeThreshold = DataSize.ofBytes(512);

    public DataSize getBufferSize() {
        return this.bufferSize;
    }

    public void setBufferSize(DataSize bufferSize) {
        this.bufferSize = bufferSize;
    }

    public DataSize getSizeThreshold() {
        return this.sizeThreshold;
    }

    public void setSizeThreshold(DataSize sizeThreshold) {
        this.sizeThreshold = sizeThreshold;
    }

}

To specify a buffer size of 10 megabytes, 10 and 10MB are equivalent. A size threshold of 256 bytes can be specified as 256 or 256B.

You can also use any of the supported units. These are:

  • B for bytes

  • KB for kilobytes

  • MB for megabytes

  • GB for gigabytes

  • TB for terabytes

The default unit is bytes and can be overridden using @DataSizeUnit as illustrated in the sample above.

If you prefer to use constructor binding, the same properties can be exposed, as shown in the following example:

@ConfigurationProperties("app.io")
@ConstructorBinding
public class AppIoProperties {

    private final DataSize bufferSize;

    private final DataSize sizeThreshold;

    public AppIoProperties(@DataSizeUnit(DataUnit.MEGABYTES) @DefaultValue("2MB") DataSize bufferSize,
            @DefaultValue("512B") DataSize sizeThreshold) {
        this.bufferSize = bufferSize;
        this.sizeThreshold = sizeThreshold;
    }

    public DataSize getBufferSize() {
        return this.bufferSize;
    }

    public DataSize getSizeThreshold() {
        return this.sizeThreshold;
    }

}
If you are upgrading a Long property, make sure to define the unit (using @DataSizeUnit) if it isn’t bytes. Doing so gives a transparent upgrade path while supporting a much richer format.

2.7.9. @ConfigurationProperties Validation

Spring Boot attempts to validate @ConfigurationProperties classes whenever they are annotated with Spring’s @Validated annotation. You can use JSR-303 javax.validation constraint annotations directly on your configuration class. To do so, ensure that a compliant JSR-303 implementation is on your classpath and then add constraint annotations to your fields, as shown in the following example:

@ConfigurationProperties(prefix="acme")
@Validated
public class AcmeProperties {

    @NotNull
    private InetAddress remoteAddress;

    // ... getters and setters

}
You can also trigger validation by annotating the @Bean method that creates the configuration properties with @Validated.

To ensure that validation is always triggered for nested properties, even when no properties are found, the associated field must be annotated with @Valid. The following example builds on the preceding AcmeProperties example:

@ConfigurationProperties(prefix="acme")
@Validated
public class AcmeProperties {

    @NotNull
    private InetAddress remoteAddress;

    @Valid
    private final Security security = new Security();

    // ... getters and setters

    public static class Security {

        @NotEmpty
        public String username;

        // ... getters and setters

    }

}

You can also add a custom Spring Validator by creating a bean definition called configurationPropertiesValidator. The @Bean method should be declared static. The configuration properties validator is created very early in the application’s lifecycle, and declaring the @Bean method as static lets the bean be created without having to instantiate the @Configuration class. Doing so avoids any problems that may be caused by early instantiation.

The spring-boot-actuator module includes an endpoint that exposes all @ConfigurationProperties beans. Point your web browser to /actuator/configprops or use the equivalent JMX endpoint. See the "Production ready features" section for details.

2.7.10. @ConfigurationProperties vs. @Value

The @Value annotation is a core container feature, and it does not provide the same features as type-safe configuration properties. The following table summarizes the features that are supported by @ConfigurationProperties and @Value:

Feature @ConfigurationProperties @Value

Relaxed binding

Yes

Limited (see note below)

Meta-data support

Yes

No

SpEL evaluation

No

Yes

If you do want to use @Value, we recommend that you refer to property names using their canonical form (kebab-case using only lowercase letters). This will allow Spring Boot to use the same logic as it does when relaxed binding @ConfigurationProperties. For example, @Value("{demo.item-price}") will pick up demo.item-price and demo.itemPrice forms from the application.properties file, as well as DEMO_ITEMPRICE from the system environment. If you used @Value("{demo.itemPrice}") instead, demo.item-price and DEMO_ITEMPRICE would not be considered.

If you define a set of configuration keys for your own components, we recommend you group them in a POJO annotated with @ConfigurationProperties. Doing so will provide you with structured, type-safe object that you can inject into your own beans.

SpEL expressions from application property files are not processed at time of parsing these files and populating the environment. However, it is possible to write a SpEL expression in @Value. If the value of a property from an application property file is a SpEL expression, it will be evaluated when consumed via @Value.

3. Profiles

Spring Profiles provide a way to segregate parts of your application configuration and make it be available only in certain environments. Any @Component, @Configuration or @ConfigurationProperties can be marked with @Profile to limit when it is loaded, as shown in the following example:

@Configuration(proxyBeanMethods = false)
@Profile("production")
public class ProductionConfiguration {

    // ...

}
If @ConfigurationProperties beans are registered via @EnableConfigurationProperties instead of automatic scanning, the @Profile annotation needs to be specified on the @Configuration class that has the @EnableConfigurationProperties annotation. In the case where @ConfigurationProperties are scanned, @Profile can be specified on the @ConfigurationProperties class itself.

You can use a spring.profiles.active Environment property to specify which profiles are active. You can specify the property in any of the ways described earlier in this chapter. For example, you could include it in your application.properties, as shown in the following example:

Properties
spring.profiles.active=dev,hsqldb
Yaml
spring:
  profiles:
    active: "dev,hsqldb"

You could also specify it on the command line by using the following switch: --spring.profiles.active=dev,hsqldb.

3.1. Adding Active Profiles

The spring.profiles.active property follows the same ordering rules as other properties: The highest PropertySource wins. This means that you can specify active profiles in application.properties and then replace them by using the command line switch.

Sometimes, it is useful to have properties that add to the active profiles rather than replace them. The SpringApplication entry point has a Java API for setting additional profiles (that is, on top of those activated by the spring.profiles.active property). See the setAdditionalProfiles() method in SpringApplication . Profile groups, which are described in the next section can also be used to add active profiles if a given profile is active.

3.2. Profile Groups

Occasionally the profiles that you define and use in your application are too fine-grained and become cumbersome to use. For example, you might have proddb and prodmq profiles that you use to enable database and messaging features independently.

To help with this, Spring Boot lets you define profile groups. A profile group allows you to define a logical name for a related group of profiles.

For example, we can create a production group that consists of our proddb and prodmq profiles.

Properties
spring.profiles.group.production[0]=proddb
spring.profiles.group.production[1]=prodmq
Yaml
spring:
  profiles:
    group:
      production:
      - "proddb"
      - "prodmq"

Our application can now be started using --spring.profiles.active=production to active the production, proddb and prodmq profiles in one hit.

3.3. Programmatically Setting Profiles

You can programmatically set active profiles by calling SpringApplication.setAdditionalProfiles(…) before your application runs. It is also possible to activate profiles by using Spring’s ConfigurableEnvironment interface.

3.4. Profile-specific Configuration Files

Profile-specific variants of both application.properties (or application.yml) and files referenced through @ConfigurationProperties are considered as files and loaded. See "Profile Specific Files" for details.

4. Logging

Spring Boot uses Commons Logging for all internal logging but leaves the underlying log implementation open. Default configurations are provided for Java Util Logging , Log4J2 , and Logback . In each case, loggers are pre-configured to use console output with optional file output also available.

By default, if you use the “Starters”, Logback is used for logging. Appropriate Logback routing is also included to ensure that dependent libraries that use Java Util Logging, Commons Logging, Log4J, or SLF4J all work correctly.

There are a lot of logging frameworks available for Java. Do not worry if the above list seems confusing. Generally, you do not need to change your logging dependencies and the Spring Boot defaults work just fine.
When you deploy your application to a servlet container or application server, logging performed via the Java Util Logging API is not routed into your application’s logs. This prevents logging performed by the container or other applications that have been deployed to it from appearing in your application’s logs.

4.1. Log Format

The default log output from Spring Boot resembles the following example:

2019-03-05 10:57:51.112  INFO 45469 --- [           main] org.apache.catalina.core.StandardEngine  : Starting Servlet Engine: Apache Tomcat/7.0.52
2019-03-05 10:57:51.253  INFO 45469 --- [ost-startStop-1] o.a.c.c.C.[Tomcat].[localhost].[/]       : Initializing Spring embedded WebApplicationContext
2019-03-05 10:57:51.253  INFO 45469 --- [ost-startStop-1] o.s.web.context.ContextLoader            : Root WebApplicationContext: initialization completed in 1358 ms
2019-03-05 10:57:51.698  INFO 45469 --- [ost-startStop-1] o.s.b.c.e.ServletRegistrationBean        : Mapping servlet: 'dispatcherServlet' to [/]
2019-03-05 10:57:51.702  INFO 45469 --- [ost-startStop-1] o.s.b.c.embedded.FilterRegistrationBean  : Mapping filter: 'hiddenHttpMethodFilter' to: [/*]

The following items are output:

  • Date and Time: Millisecond precision and easily sortable.

  • Log Level: ERROR, WARN, INFO, DEBUG, or TRACE.

  • Process ID.

  • A --- separator to distinguish the start of actual log messages.

  • Thread name: Enclosed in square brackets (may be truncated for console output).

  • Logger name: This is usually the source class name (often abbreviated).

  • The log message.

Logback does not have a FATAL level. It is mapped to ERROR.

4.2. Console Output

The default log configuration echoes messages to the console as they are written. By default, ERROR-level, WARN-level, and INFO-level messages are logged. You can also enable a “debug” mode by starting your application with a --debug flag.

$ java -jar myapp.jar --debug
You can also specify debug=true in your application.properties.

When the debug mode is enabled, a selection of core loggers (embedded container, Hibernate, and Spring Boot) are configured to output more information. Enabling the debug mode does not configure your application to log all messages with DEBUG level.

Alternatively, you can enable a “trace” mode by starting your application with a --trace flag (or trace=true in your application.properties). Doing so enables trace logging for a selection of core loggers (embedded container, Hibernate schema generation, and the whole Spring portfolio).

4.2.1. Color-coded Output

If your terminal supports ANSI, color output is used to aid readability. You can set spring.output.ansi.enabled to a supported value to override the auto-detection.

Color coding is configured by using the %clr conversion word. In its simplest form, the converter colors the output according to the log level, as shown in the following example:

%clr(%5p)

The following table describes the mapping of log levels to colors:

Level Color

FATAL

Red

ERROR

Red

WARN

Yellow

INFO

Green

DEBUG

Green

TRACE

Green

Alternatively, you can specify the color or style that should be used by providing it as an option to the conversion. For example, to make the text yellow, use the following setting:

%clr(%d{yyyy-MM-dd HH:mm:ss.SSS}){yellow}

The following colors and styles are supported:

  • blue

  • cyan

  • faint

  • green

  • magenta

  • red

  • yellow

4.3. File Output

By default, Spring Boot logs only to the console and does not write log files. If you want to write log files in addition to the console output, you need to set a logging.file.name or logging.file.path property (for example, in your application.properties).

The following table shows how the logging.* properties can be used together:

Table 5. Logging properties
logging.file.name logging.file.path Example Description

(none)

(none)

Console only logging.

Specific file

(none)

my.log

Writes to the specified log file. Names can be an exact location or relative to the current directory.

(none)

Specific directory

/var/log

Writes spring.log to the specified directory. Names can be an exact location or relative to the current directory.

Log files rotate when they reach 10 MB and, as with console output, ERROR-level, WARN-level, and INFO-level messages are logged by default.

Logging properties are independent of the actual logging infrastructure. As a result, specific configuration keys (such as logback.configurationFile for Logback) are not managed by spring Boot.

4.4. File Rotation

If you are using the Logback, it’s possible to fine-tune log rotation settings using your application.properties or application.yaml file. For all other logging system, you’ll need to configure rotation settings directly yourself (for example, if you use Log4J2 then you could add a log4j.xml file).

The following rotation policy properties are supported:

Name Description

logging.logback.rollingpolicy.file-name-pattern

The filename pattern used to create log archives.

logging.logback.rollingpolicy.clean-history-on-start

If log archive cleanup should occur when the application starts.

logging.logback.rollingpolicy.max-file-size

The maximum size of log file before it’s archived.

logging.logback.rollingpolicy.total-size-cap

The maximum amount of size log archives can take before being deleted.

logging.logback.rollingpolicy.max-history

The number of days to keep log archives (defaults to 7)

4.5. Log Levels

All the supported logging systems can have the logger levels set in the Spring Environment (for example, in application.properties) by using logging.level.<logger-name>=<level> where level is one of TRACE, DEBUG, INFO, WARN, ERROR, FATAL, or OFF. The root logger can be configured by using logging.level.root.

The following example shows potential logging settings in application.properties:

Properties
logging.level.root=warn
logging.level.org.springframework.web=debug
logging.level.org.hibernate=error
Yaml
logging:
  level:
    root: "warn"
    org.springframework.web: "debug"
    org.hibernate: "error"

It’s also possible to set logging levels using environment variables. For example, LOGGING_LEVEL_ORG_SPRINGFRAMEWORK_WEB=DEBUG will set org.springframework.web to DEBUG.

The above approach will only work for package level logging. Since relaxed binding always converts environment variables to lowercase, it’s not possible to configure logging for an individual class in this way. If you need to configure logging for a class, you can use the SPRING_APPLICATION_JSON variable.

4.6. Log Groups

It’s often useful to be able to group related loggers together so that they can all be configured at the same time. For example, you might commonly change the logging levels for all Tomcat related loggers, but you can’t easily remember top level packages.

To help with this, Spring Boot allows you to define logging groups in your Spring Environment. For example, here’s how you could define a “tomcat” group by adding it to your application.properties:

Properties
logging.group.tomcat=org.apache.catalina,org.apache.coyote,org.apache.tomcat
Yaml
logging:
  group:
    tomcat: "org.apache.catalina,org.apache.coyote,org.apache.tomcat"

Once defined, you can change the level for all the loggers in the group with a single line:

Properties
logging.level.tomcat=trace
Yaml
logging:
  level:
    tomcat: "trace"

Spring Boot includes the following pre-defined logging groups that can be used out-of-the-box:

Name Loggers

web

org.springframework.core.codec, org.springframework.http, org.springframework.web, org.springframework.boot.actuate.endpoint.web, org.springframework.boot.web.servlet.ServletContextInitializerBeans

sql

org.springframework.jdbc.core, org.hibernate.SQL, org.jooq.tools.LoggerListener

4.7. Using a Log Shutdown Hook

In order to release logging resources it is usually a good idea to stop the logging system when your application terminates. Unfortunately, there’s no single way to do this that will work with all application types. If your application has complex context hierarchies or is deployed as a war file, you’ll need to investigate the options provided directly by the underlying logging system. For example, Logback offers context selectors which allow each Logger to be created in its own context.

For simple "single jar" applications deployed in their own JVM, you can use the logging.register-shutdown-hook property. Setting logging.register-shutdown-hook to true will register a shutdown hook that will trigger log system cleanup when the JVM exits.

You can set the property in your application.properties or application.yaml file:

Properties
logging.register-shutdown-hook=true
Yaml
logging:
  register-shutdown-hook: true

4.8. Custom Log Configuration

The various logging systems can be activated by including the appropriate libraries on the classpath and can be further customized by providing a suitable configuration file in the root of the classpath or in a location specified by the following Spring Environment property: logging.config.

You can force Spring Boot to use a particular logging system by using the org.springframework.boot.logging.LoggingSystem system property. The value should be the fully qualified class name of a LoggingSystem implementation. You can also disable Spring Boot’s logging configuration entirely by using a value of none.

Since logging is initialized before the ApplicationContext is created, it is not possible to control logging from @PropertySources in Spring @Configuration files. The only way to change the logging system or disable it entirely is via System properties.

Depending on your logging system, the following files are loaded:

Logging System Customization

Logback

logback-spring.xml, logback-spring.groovy, logback.xml, or logback.groovy

Log4j2

log4j2-spring.xml or log4j2.xml

JDK (Java Util Logging)

logging.properties

When possible, we recommend that you use the -spring variants for your logging configuration (for example, logback-spring.xml rather than logback.xml). If you use standard configuration locations, Spring cannot completely control log initialization.
There are known classloading issues with Java Util Logging that cause problems when running from an 'executable jar'. We recommend that you avoid it when running from an 'executable jar' if at all possible.

To help with the customization, some other properties are transferred from the Spring Environment to System properties, as described in the following tab