“How-to” Guides

FailureAnalyzer is a great way to intercept an exception on startup and turn it into a human-readable message, wrapped in a FailureAnalysis. Spring Boot provides such an analyzer for application-context-related exceptions, JSR-303 validations, and more. You can also create your own.

AbstractFailureAnalyzer is a convenient extension of FailureAnalyzer that checks the presence of a specified exception type in the exception to handle. You can extend from that so that your implementation gets a chance to handle the exception only when it is actually present. If, for whatever reason, you cannot handle the exception, return null to give another implementation a chance to handle the exception.

FailureAnalyzer implementations must be registered in META-INF/spring.factories. The following example registers ProjectConstraintViolationFailureAnalyzer:

The Spring Boot auto-configuration tries its best to “do the right thing”, but sometimes things fail, and it can be hard to tell why.

There is a really useful ConditionEvaluationReport available in any Spring Boot ApplicationContext. You can see it if you enable DEBUG logging output. If you use the spring-boot-actuator (see the Actuator chapter), there is also a conditions endpoint that renders the report in JSON. Use that endpoint to debug the application and see what features have been added (and which have not been added) by Spring Boot at runtime.

Many more questions can be answered by looking at the source code and the Javadoc. When reading the code, remember the following rules of thumb:

A SpringApplication has ApplicationListeners and ApplicationContextInitializers that are used to apply customizations to the context or environment. Spring Boot loads a number of such customizations for use internally from META-INF/spring.factories. There is more than one way to register additional customizations:

The SpringApplication sends some special ApplicationEvents to the listeners (some even before the context is created) and then registers the listeners for events published by the ApplicationContext as well. See “Application Events and Listeners” in the ‘Spring Boot features’ section for a complete list.

It is also possible to customize the Environment before the application context is refreshed by using EnvironmentPostProcessor. Each implementation should be registered in META-INF/spring.factories, as shown in the following example:

The implementation can load arbitrary files and add them to the Environment. For instance, the following example loads a YAML configuration file from the classpath:

You can use the ApplicationBuilder class to create parent/child ApplicationContext hierarchies. See “spring-boot-features.html” in the ‘Spring Boot features’ section for more information.

Not all Spring applications have to be web applications (or web services). If you want to execute some code in a main method but also bootstrap a Spring application to set up the infrastructure to use, you can use the SpringApplication features of Spring Boot. A SpringApplication changes its ApplicationContext class, depending on whether it thinks it needs a web application or not. The first thing you can do to help it is to leave server-related dependencies (e.g. servlet API) off the classpath. If you cannot do that (for example, you run two applications from the same code base) then you can explicitly call setWebApplicationType(WebApplicationType.NONE) on your SpringApplication instance or set the applicationContextClass property (through the Java API or with external properties). Application code that you want to run as your business logic can be implemented as a CommandLineRunner and dropped into the context as a @Bean definition.

Rather than hardcoding some properties that are also specified in your project’s build configuration, you can automatically expand them by instead using the existing build configuration. This is possible in both Maven and Gradle.

A SpringApplication has bean properties (mainly setters), so you can use its Java API as you create the application to modify its behavior. Alternatively, you can externalize the configuration by setting properties in spring.main.*. For example, in application.properties, you might have the following settings:

Then the Spring Boot banner is not printed on startup, and the application is not starting an embedded web server.

Properties defined in external configuration override the values specified with the Java API, with the notable exception of the sources used to create the ApplicationContext. Consider the following application:

Now consider the following configuration:

The actual application now shows the banner (as overridden by configuration) and uses three sources for the ApplicationContext (in the following order): demo.MyApp, com.acme.Config, and com.acme.ExtraConfig.

By default, properties from different sources are added to the Spring Environment in a defined order (see “spring-boot-features.html” in the ‘Spring Boot features’ section for the exact order).

You can also provide the following System properties (or environment variables) to change the behavior:

No matter what you set in the environment, Spring Boot always loads application.properties as described above. By default, if YAML is used, then files with the ‘.yml’ extension are also added to the list.

Spring Boot logs the configuration files that are loaded at the DEBUG level and the candidates it has not found at TRACE level.

See ConfigFileApplicationListener for more detail.

Some people like to use (for example) --port=9000 instead of --server.port=9000 to set configuration properties on the command line. You can enable this behavior by using placeholders in application.properties, as shown in the following example:

YAML is a superset of JSON and, as such, is a convenient syntax for storing external properties in a hierarchical format, as shown in the following example:

Create a file called application.yml and put it in the root of your classpath. Then add snakeyaml to your dependencies (Maven coordinates org.yaml:snakeyaml, already included if you use the spring-boot-starter). A YAML file is parsed to a Java Map<String,Object> (like a JSON object), and Spring Boot flattens the map so that it is one level deep and has period-separated keys, as many people are used to with Properties files in Java.

The preceding example YAML corresponds to the following application.properties file:

See “spring-boot-features.html” in the ‘Spring Boot features’ section for more information about YAML.

The Spring Environment has an API for this, but you would normally set a System property (spring.profiles.active) or an OS environment variable (SPRING_PROFILES_ACTIVE). Also, you can launch your application with a -D argument (remember to put it before the main class or jar archive), as follows:

In Spring Boot, you can also set the active profile in application.properties, as shown in the following example:

A value set this way is replaced by the System property or environment variable setting but not by the SpringApplicationBuilder.profiles() method. Thus, the latter Java API can be used to augment the profiles without changing the defaults.

See “spring-boot-features.html” in the “Spring Boot features” section for more information.

A YAML file is actually a sequence of documents separated by --- lines, and each document is parsed separately to a flattened map.

If a YAML document contains a spring.profiles key, then the profiles value (a comma-separated list of profiles) is fed into the Spring Environment.acceptsProfiles() method. If any of those profiles is active, that document is included in the final merge (otherwise, it is not), as shown in the following example:

In the preceding example, the default port is 9000. However, if the Spring profile called ‘development’ is active, then the port is 9001. If ‘production’ is active, then the port is 0.

To do the same thing with properties files, you can use application-${profile}.properties to specify profile-specific values.

Spring Boot binds external properties from application.properties (or .yml files and other places) into an application at runtime. There is not (and technically cannot be) an exhaustive list of all supported properties in a single location, because contributions can come from additional jar files on your classpath.

A running application with the Actuator features has a configprops endpoint that shows all the bound and bindable properties available through @ConfigurationProperties.

The appendix includes an application.properties example with a list of the most common properties supported by Spring Boot. The definitive list comes from searching the source code for @ConfigurationProperties and @Value annotations as well as the occasional use of Binder. For more about the exact ordering of loading properties, see "spring-boot-features.html".

Many Spring Boot starters include default embedded containers.

When switching to a different HTTP server, you need to swap the default dependencies for those that you need instead. To help with this process, Spring Boot provides a separate starter for each of the supported HTTP servers.

The following Maven example shows how to exclude Tomcat and include Jetty for Spring MVC:

The following Gradle example configures the necessary dependencies and a module replacement to use Undertow in place of Reactor Netty for Spring WebFlux:

If your classpath contains the necessary bits to start a web server, Spring Boot will automatically start it. To disable this behavior configure the WebApplicationType in your application.properties, as shown in the following example:

In a standalone application, the main HTTP port defaults to 8080 but can be set with server.port (for example, in application.properties or as a System property). Thanks to relaxed binding of Environment values, you can also use SERVER_PORT (for example, as an OS environment variable).

To switch off the HTTP endpoints completely but still create a WebApplicationContext, use server.port=-1 (doing so is sometimes useful for testing).

For more details, see “spring-boot-features.html” in the ‘Spring Boot Features’ section, or the ServerProperties source code.

To scan for a free port (using OS natives to prevent clashes) use server.port=0.

You can access the port the server is running on from log output or from the WebServerApplicationContext through its WebServer. The best way to get that and be sure it has been initialized is to add a @Bean of type ApplicationListener<WebServerInitializedEvent> and pull the container out of the event when it is published.

Tests that use @SpringBootTest(webEnvironment=WebEnvironment.RANDOM_PORT) can also inject the actual port into a field by using the @LocalServerPort annotation, as shown in the following example:

HTTP response compression is supported by Jetty, Tomcat, and Undertow. It can be enabled in application.properties, as follows:

By default, responses must be at least 2048 bytes in length for compression to be performed. You can configure this behavior by setting the server.compression.min-response-size property.

By default, responses are compressed only if their content type is one of the following:

You can configure this behavior by setting the server.compression.mime-types property.

SSL can be configured declaratively by setting the various server.ssl.* properties, typically in application.properties or application.yml. The following example shows setting SSL properties in application.properties:

See Ssl for details of all of the supported properties.

Using configuration such as the preceding example means the application no longer supports a plain HTTP connector at port 8080. Spring Boot does not support the configuration of both an HTTP connector and an HTTPS connector through application.properties. If you want to have both, you need to configure one of them programmatically. We recommend using application.properties to configure HTTPS, as the HTTP connector is the easier of the two to configure programmatically.

You can enable HTTP/2 support in your Spring Boot application with the server.http2.enabled configuration property. This support depends on the chosen web server and the application environment, since that protocol is not supported out-of-the-box by all JDK8 releases.

Generally, you should first consider using one of the many available configuration keys and customize your web server by adding new entries in your application.properties (or application.yml, or environment, etc. see “Discover Built-in Options for External Properties”). The server.* namespace is quite useful here, and it includes namespaces like server.tomcat.*, server.jetty.* and others, for server-specific features. See the list of appendix-application-properties.html.

The previous sections covered already many common use cases, such as compression, SSL or HTTP/2. However, if a configuration key doesn’t exist for your use case, you should then look at WebServerFactoryCustomizer. You can declare such a component and get access to the server factory relevant to your choice: you should select the variant for the chosen Server (Tomcat, Jetty, Reactor Netty, Undertow) and the chosen web stack (Servlet or Reactive).

The example below is for Tomcat with the spring-boot-starter-web (Servlet stack):

Once you’ve got access to a WebServerFactory using the customizer, you can use it to configure specific parts, like connectors, server resources, or the server itself - all using server-specific APIs.

In addition Spring Boot provides:

As a last resort, you can also declare your own WebServerFactory bean, which will override the one provided by Spring Boot. When you do so, auto-configured customizers are still applied on your custom factory, so use that option carefully.

In a servlet stack application, i.e. with the spring-boot-starter-web, there are two ways to add Servlet, Filter, ServletContextListener, and the other listeners supported by the Servlet API to your application:

Access logs can be configured for Tomcat, Undertow, and Jetty through their respective namespaces.

For instance, the following settings log access on Tomcat with a custom pattern.

Access logging for Undertow can be configured in a similar fashion, as shown in the following example:

Logs are stored in a logs directory relative to the working directory of the application. You can customize this location by setting the server.undertow.accesslog.dir property.

Finally, access logging for Jetty can also be configured as follows:

By default, logs are redirected to System.err. For more details, see the Jetty documentation.

If your application is running behind a proxy, a load-balancer or in the cloud, the request information (like the host, port, scheme…​) might change along the way. Your application may be running on 10.10.10.10:8080, but HTTP clients should only see example.org.

RFC7239 "Forwarded Headers" defines the Forwarded HTTP header; proxies can use this header to provide information about the original request. You can configure your application to read those headers and automatically use that information when creating links and sending them to clients in HTTP 302 responses, JSON documents or HTML pages. There are also non-standard headers, like X-Forwarded-Host, X-Forwarded-Port, X-Forwarded-Proto, X-Forwarded-Ssl, and X-Forwarded-Prefix.

If the proxy adds the commonly used X-Forwarded-For and X-Forwarded-Proto headers, setting server.forward-headers-strategy to NATIVE is enough to support those. With this option, the Web servers themselves natively support this feature; you can check their specific documentation to learn about specific behavior.

If this is not enough, Spring Framework provides a ForwardedHeaderFilter. You can register it as a Servlet Filter in your application by setting server.forward-headers-strategy is set to FRAMEWORK.

You can add an org.apache.catalina.connector.Connector to the TomcatServletWebServerFactory, which can allow multiple connectors, including HTTP and HTTPS connectors, as shown in the following example:

By default, the embedded Tomcat used by Spring Boot does not support "Version 0" of the Cookie format, so you may see the following error:

If at all possible, you should consider updating your code to only store values compliant with later Cookie specifications. If, however, you cannot change the way that cookies are written, you can instead configure Tomcat to use a LegacyCookieProcessor. To switch to the LegacyCookieProcessor, use an WebServerFactoryCustomizer bean that adds a TomcatContextCustomizer, as shown in the following example:

Embedded Tomcat’s MBean registry is disabled by default. This minimizes Tomcat’s memory footprint. If you want to use Tomcat’s MBeans, for example so that they can be used to expose metrics via Micrometer, you must use the server.tomcat.mbeanregistry.enabled property to do so, as shown in the following example:

Add an UndertowBuilderCustomizer to the UndertowServletWebServerFactory and add a listener to the Builder, as shown in the following example:

If you want to use @ServerEndpoint in a Spring Boot application that used an embedded container, you must declare a single ServerEndpointExporter @Bean, as shown in the following example:

The bean shown in the preceding example registers any @ServerEndpoint annotated beans with the underlying WebSocket container. When deployed to a standalone servlet container, this role is performed by a servlet container initializer, and the ServerEndpointExporter bean is not required.

Any Spring @RestController in a Spring Boot application should render JSON response by default as long as Jackson2 is on the classpath, as shown in the following example:

As long as MyThing can be serialized by Jackson2 (true for a normal POJO or Groovy object), then localhost:8080/thing serves a JSON representation of it by default. Note that, in a browser, you might sometimes see XML responses, because browsers tend to send accept headers that prefer XML.

If you have the Jackson XML extension (jackson-dataformat-xml) on the classpath, you can use it to render XML responses. The previous example that we used for JSON would work. To use the Jackson XML renderer, add the following dependency to your project:

If Jackson’s XML extension is not available and JAXB is available, XML can be rendered with the additional requirement of having MyThing annotated as @XmlRootElement, as shown in the following example:

JAXB is only available out of the box with Java 8. If you’re using a more recent Java generation, add the following dependency to your project:

Spring MVC (client and server side) uses HttpMessageConverters to negotiate content conversion in an HTTP exchange. If Jackson is on the classpath, you already get the default converter(s) provided by Jackson2ObjectMapperBuilder, an instance of which is auto-configured for you.

The ObjectMapper (or XmlMapper for Jackson XML converter) instance (created by default) has the following customized properties:

Spring Boot also has some features to make it easier to customize this behavior.

You can configure the ObjectMapper and XmlMapper instances by using the environment. Jackson provides an extensive suite of on/off features that can be used to configure various aspects of its processing. These features are described in six enums (in Jackson) that map onto properties in the environment:

For example, to enable pretty print, set spring.jackson.serialization.indent_output=true. Note that, thanks to the use of relaxed binding, the case of indent_output does not have to match the case of the corresponding enum constant, which is INDENT_OUTPUT.

This environment-based configuration is applied to the auto-configured Jackson2ObjectMapperBuilder bean and applies to any mappers created by using the builder, including the auto-configured ObjectMapper bean.

The context’s Jackson2ObjectMapperBuilder can be customized by one or more Jackson2ObjectMapperBuilderCustomizer beans. Such customizer beans can be ordered (Boot’s own customizer has an order of 0), letting additional customization be applied both before and after Boot’s customization.

Any beans of type com.fasterxml.jackson.databind.Module are automatically registered with the auto-configured Jackson2ObjectMapperBuilder and are applied to any ObjectMapper instances that it creates. This provides a global mechanism for contributing custom modules when you add new features to your application.

If you want to replace the default ObjectMapper completely, either define a @Bean of that type and mark it as @Primary or, if you prefer the builder-based approach, define a Jackson2ObjectMapperBuilder @Bean. Note that, in either case, doing so disables all auto-configuration of the ObjectMapper.

If you provide any @Beans of type MappingJackson2HttpMessageConverter, they replace the default value in the MVC configuration. Also, a convenience bean of type HttpMessageConverters is provided (and is always available if you use the default MVC configuration). It has some useful methods to access the default and user-enhanced message converters.

See the “Customize the @ResponseBody Rendering” section and the WebMvcAutoConfiguration source code for more details.

Spring uses HttpMessageConverters to render @ResponseBody (or responses from @RestController). You can contribute additional converters by adding beans of the appropriate type in a Spring Boot context. If a bean you add is of a type that would have been included by default anyway (such as MappingJackson2HttpMessageConverter for JSON conversions), it replaces the default value. A convenience bean of type HttpMessageConverters is provided and is always available if you use the default MVC configuration. It has some useful methods to access the default and user-enhanced message converters (For example, it can be useful if you want to manually inject them into a custom RestTemplate).

As in normal MVC usage, any WebMvcConfigurer beans that you provide can also contribute converters by overriding the configureMessageConverters method. However, unlike with normal MVC, you can supply only additional converters that you need (because Spring Boot uses the same mechanism to contribute its defaults). Finally, if you opt out of the Spring Boot default MVC configuration by providing your own @EnableWebMvc configuration, you can take control completely and do everything manually by using getMessageConverters from WebMvcConfigurationSupport.

See the WebMvcAutoConfiguration source code for more details.

Spring Boot embraces the Servlet 3 javax.servlet.http.Part API to support uploading files. By default, Spring Boot configures Spring MVC with a maximum size of 1MB per file and a maximum of 10MB of file data in a single request. You may override these values, the location to which intermediate data is stored (for example, to the /tmp directory), and the threshold past which data is flushed to disk by using the properties exposed in the MultipartProperties class. For example, if you want to specify that files be unlimited, set the spring.servlet.multipart.max-file-size property to -1.

The multipart support is helpful when you want to receive multipart encoded file data as a @RequestParam-annotated parameter of type MultipartFile in a Spring MVC controller handler method.

See the MultipartAutoConfiguration source for more details.

By default, all content is served from the root of your application (/). If you would rather map to a different path, you can configure one as follows:

If you have additional servlets you can declare a @Bean of type Servlet or ServletRegistrationBean for each and Spring Boot will register them transparently to the container. Because servlets are registered that way, they can be mapped to a sub-context of the DispatcherServlet without invoking it.

Configuring the DispatcherServlet yourself is unusual but if you really need to do it, a @Bean of type DispatcherServletPath must be provided as well to provide the path of your custom DispatcherServlet.

The easiest way to take complete control over MVC configuration is to provide your own @Configuration with the @EnableWebMvc annotation. Doing so leaves all MVC configuration in your hands.

A ViewResolver is a core component of Spring MVC, translating view names in @Controller to actual View implementations. Note that ViewResolvers are mainly used in UI applications, rather than REST-style services (a View is not used to render a @ResponseBody). There are many implementations of ViewResolver to choose from, and Spring on its own is not opinionated about which ones you should use. Spring Boot, on the other hand, installs one or two for you, depending on what it finds on the classpath and in the application context. The DispatcherServlet uses all the resolvers it finds in the application context, trying each one in turn until it gets a result. If you add your own, you have to be aware of the order and in which position your resolver is added.

WebMvcAutoConfiguration adds the following ViewResolvers to your context:

For more detail, see the following sections:

Spring Security can be used to secure a Jersey-based web application in much the same way as it can be used to secure a Spring MVC-based web application. However, if you want to use Spring Security’s method-level security with Jersey, you must configure Jersey to use setStatus(int) rather sendError(int). This prevents Jersey from committing the response before Spring Security has had an opportunity to report an authentication or authorization failure to the client.

The jersey.config.server.response.setStatusOverSendError property must be set to true on the application’s ResourceConfig bean, as shown in the following example:

To use Jersey alongside another web framework, such as Spring MVC, it should be configured so that it will allow the other framework to handle requests that it cannot handle. First, configure Jersey to use a Filter rather than a Servlet by configuring the spring.jersey.type application property with a value of filter. Second, configure your ResourceConfig to forward requests that would have resulted in a 404, as shown in the following example.

As described in spring-boot-features.html, you can use a RestTemplateCustomizer with RestTemplateBuilder to build a customized RestTemplate. This is the recommended approach for creating a RestTemplate configured to use a proxy.

The exact details of the proxy configuration depend on the underlying client request factory that is being used. The following example configures HttpComponentsClientRequestFactory with an HttpClient that uses a proxy for all hosts except 192.168.0.5:

When Reactor Netty is on the classpath a Reactor Netty-based WebClient is auto-configured. To customize the client’s handling of network connections, provide a ClientHttpConnector bean. The following example configures a 60 second connect timeout and adds a ReadTimeoutHandler:

If you need to apply customizations to logback beyond those that can be achieved with application.properties, you’ll need to add a standard logback configuration file. You can add a logback.xml file to the root of your classpath for logback to find. You can also use logback-spring.xml if you want to use the Spring Boot Logback extensions.

Spring Boot provides a number of logback configurations that be included from your own configuration. These includes are designed to allow certain common Spring Boot conventions to be re-applied.

The following files are provided under org/springframework/boot/logging/logback/:

In addition, a legacy base.xml file is provided for compatibility with earlier versions of Spring Boot.

A typical custom logback.xml file would look something like this:

Your logback configuration file can also make use of System properties that the LoggingSystem takes care of creating for you:

Spring Boot also provides some nice ANSI color terminal output on a console (but not in a log file) by using a custom Logback converter. See the CONSOLE_LOG_PATTERN in the defaults.xml configuration for an example.

If Groovy is on the classpath, you should be able to configure Logback with logback.groovy as well. If present, this setting is given preference.

Spring Boot supports Log4j 2 for logging configuration if it is on the classpath. If you use the starters for assembling dependencies, you have to exclude Logback and then include log4j 2 instead. If you do not use the starters, you need to provide (at least) spring-jcl in addition to Log4j 2.

The recommended path is through the starters, even though it requires some jiggling. The following example shows how to set up the starters in Maven:

Gradle provides a few different ways to set up the starters. One way is to use a module replacement. To do so, declare a dependency on the Log4j 2 starter and tell Gradle that any occurrences of the default logging starter should be replaced by the Log4j 2 starter, as shown in the following example:

To configure your own DataSource, define a @Bean of that type in your configuration. Spring Boot reuses your DataSource anywhere one is required, including database initialization. If you need to externalize some settings, you can bind your DataSource to the environment (see “spring-boot-features.html”).

The following example shows how to define a data source in a bean:

The following example shows how to define a data source by setting properties:

Assuming that your FancyDataSource has regular JavaBean properties for the URL, the username, and the pool size, these settings are bound automatically before the DataSource is made available to other components. The regular database initialization also happens (so the relevant sub-set of spring.datasource.* can still be used with your custom configuration).

Spring Boot also provides a utility builder class, called DataSourceBuilder, that can be used to create one of the standard data sources (if it is on the classpath). The builder can detect the one to use based on what’s available on the classpath. It also auto-detects the driver based on the JDBC URL.

The following example shows how to create a data source by using a DataSourceBuilder:

To run an app with that DataSource, all you need is the connection information. Pool-specific settings can also be provided. Check the implementation that is going to be used at runtime for more details.

The following example shows how to define a JDBC data source by setting properties:

However, there is a catch. Because the actual type of the connection pool is not exposed, no keys are generated in the metadata for your custom DataSource and no completion is available in your IDE (because the DataSource interface exposes no properties). Also, if you happen to have Hikari on the classpath, this basic setup does not work, because Hikari has no url property (but does have a jdbcUrl property). In that case, you must rewrite your configuration as follows:

You can fix that by forcing the connection pool to use and return a dedicated implementation rather than DataSource. You cannot change the implementation at runtime, but the list of options will be explicit.

The following example shows how create a HikariDataSource with DataSourceBuilder:

You can even go further by leveraging what DataSourceProperties does for you — that is, by providing a default embedded database with a sensible username and password if no URL is provided. You can easily initialize a DataSourceBuilder from the state of any DataSourceProperties object, so you could also inject the DataSource that Spring Boot creates automatically. However, that would split your configuration into two namespaces: url, username, password, type, and driver on spring.datasource and the rest on your custom namespace (app.datasource). To avoid that, you can redefine a custom DataSourceProperties on your custom namespace, as shown in the following example:

This setup puts you in sync with what Spring Boot does for you by default, except that a dedicated connection pool is chosen (in code) and its settings are exposed in the app.datasource.configuration sub namespace. Because DataSourceProperties is taking care of the url/jdbcUrl translation for you, you can configure it as follows:

See “spring-boot-features.html” in the “Spring Boot features” section and the DataSourceAutoConfiguration class for more details.

If you need to configure multiple data sources, you can apply the same tricks that are described in the previous section. You must, however, mark one of the DataSource instances as @Primary, because various auto-configurations down the road expect to be able to get one by type.

If you create your own DataSource, the auto-configuration backs off. In the following example, we provide the exact same feature set as the auto-configuration provides on the primary data source:

Both data sources are also bound for advanced customizations. For instance, you could configure them as follows:

You can apply the same concept to the secondary DataSource as well, as shown in the following example:

The preceding example configures two data sources on custom namespaces with the same logic as Spring Boot would use in auto-configuration. Note that each configuration sub namespace provides advanced settings based on the chosen implementation.

Spring Data can create implementations of @Repository interfaces of various flavors. Spring Boot handles all of that for you, as long as those @Repositories are included in the same package (or a sub-package) of your @EnableAutoConfiguration class.

For many applications, all you need is to put the right Spring Data dependencies on your classpath. There is a spring-boot-starter-data-jpa for JPA, spring-boot-starter-data-mongodb for Mongodb, etc. To get started, create some repository interfaces to handle your @Entity objects.

Spring Boot tries to guess the location of your @Repository definitions, based on the @EnableAutoConfiguration it finds. To get more control, use the @EnableJpaRepositories annotation (from Spring Data JPA).

For more about Spring Data, see the Spring Data project page.

Spring Boot tries to guess the location of your @Entity definitions, based on the @EnableAutoConfiguration it finds. To get more control, you can use the @EntityScan annotation, as shown in the following example:

Spring Data JPA already provides some vendor-independent configuration options (such as those for SQL logging), and Spring Boot exposes those options and a few more for Hibernate as external configuration properties. Some of them are automatically detected according to the context so you should not have to set them.

The spring.jpa.hibernate.ddl-auto is a special case, because, depending on runtime conditions, it has different defaults. If an embedded database is used and no schema manager (such as Liquibase or Flyway) is handling the DataSource, it defaults to create-drop. In all other cases, it defaults to none.

The dialect to use is detected by the JPA provider. If you prefer to set the dialect yourself, set the spring.jpa.database-platform property.

The most common options to set are shown in the following example:

In addition, all properties in spring.jpa.properties.* are passed through as normal JPA properties (with the prefix stripped) when the local EntityManagerFactory is created.

Hibernate uses two different naming strategies to map names from the object model to the corresponding database names. The fully qualified class name of the physical and the implicit strategy implementations can be configured by setting the spring.jpa.hibernate.naming.physical-strategy and spring.jpa.hibernate.naming.implicit-strategy properties, respectively. Alternatively, if ImplicitNamingStrategy or PhysicalNamingStrategy beans are available in the application context, Hibernate will be automatically configured to use them.

By default, Spring Boot configures the physical naming strategy with SpringPhysicalNamingStrategy. This implementation provides the same table structure as Hibernate 4: all dots are replaced by underscores and camel casing is replaced by underscores as well. Additionally, by default, all table names are generated in lower case. For example, a TelephoneNumber entity is mapped to the telephone_number table. If your schema requires mixed-case identifiers, define a custom SpringPhysicalNamingStrategy bean, as shown in the following example:

If you prefer to use Hibernate 5’s default instead, set the following property:

Alternatively, you can configure the following bean:

See HibernateJpaAutoConfiguration and JpaBaseConfiguration for more details.

Hibernate second-level cache can be configured for a range of cache providers. Rather than configuring Hibernate to lookup the cache provider again, it is better to provide the one that is available in the context whenever possible.

To do this with JCache, first make sure that org.hibernate:hibernate-jcache is available on the classpath. Then, add a HibernatePropertiesCustomizer bean as shown in the following example:

This customizer will configure Hibernate to use the same CacheManager as the one that the application uses. It is also possible to use separate CacheManager instances. For details, refer to the Hibernate user guide.

By default, Spring Boot registers a BeanContainer implementation that uses the BeanFactory so that converters and entity listeners can use regular dependency injection.

You can disable or tune this behaviour by registering a HibernatePropertiesCustomizer that removes or changes the hibernate.resource.beans.container property.

To take full control of the configuration of the EntityManagerFactory, you need to add a @Bean named ‘entityManagerFactory’. Spring Boot auto-configuration switches off its entity manager in the presence of a bean of that type.

If you need to use JPA against multiple data sources, you likely need one EntityManagerFactory per data source. The LocalContainerEntityManagerFactoryBean from Spring ORM allows you to configure an EntityManagerFactory for your needs. You can also reuse JpaProperties to bind settings for each EntityManagerFactory, as shown in the following example:

The example above creates an EntityManagerFactory using a DataSource bean named firstDataSource. It scans entities located in the same package as Order. It is possible to map additional JPA properties using the app.first.jpa namespace.

You should provide a similar configuration for any additional data sources for which you need JPA access. To complete the picture, you need to configure a JpaTransactionManager for each EntityManagerFactory as well. Alternatively, you might be able to use a JTA transaction manager that spans both.

If you use Spring Data, you need to configure @EnableJpaRepositories accordingly, as shown in the following example:

Spring Boot will not search for or use a META-INF/persistence.xml by default. If you prefer to use a traditional persistence.xml, you need to define your own @Bean of type LocalEntityManagerFactoryBean (with an ID of ‘entityManagerFactory’) and set the persistence unit name there.

See JpaBaseConfiguration for the default settings.

Spring Data JPA and Spring Data Mongo can both automatically create Repository implementations for you. If they are both present on the classpath, you might have to do some extra configuration to tell Spring Boot which repositories to create. The most explicit way to do that is to use the standard Spring Data @EnableJpaRepositories and @EnableMongoRepositories annotations and provide the location of your Repository interfaces.

There are also flags (spring.data.*.repositories.enabled and spring.data.*.repositories.type) that you can use to switch the auto-configured repositories on and off in external configuration. Doing so is useful, for instance, in case you want to switch off the Mongo repositories and still use the auto-configured MongoTemplate.

The same obstacle and the same features exist for other auto-configured Spring Data repository types (Elasticsearch, Solr, and others). To work with them, change the names of the annotations and flags accordingly.

Spring Data provides web support that simplifies the use of Spring Data repositories in a web application. Spring Boot provides properties in the spring.data.web namespace for customizing its configuration. Note that if you are using Spring Data REST, you must use the properties in the spring.data.rest namespace instead.

Spring Data REST can expose the Repository implementations as REST endpoints for you, provided Spring MVC has been enabled for the application.

Spring Boot exposes a set of useful properties (from the spring.data.rest namespace) that customize the RepositoryRestConfiguration. If you need to provide additional customization, you should use a RepositoryRestConfigurer bean.

If you want to configure a component that JPA uses, then you need to ensure that the component is initialized before JPA. When the component is auto-configured, Spring Boot takes care of this for you. For example, when Flyway is auto-configured, Hibernate is configured to depend upon Flyway so that Flyway has a chance to initialize the database before Hibernate tries to use it.

If you are configuring a component yourself, you can use an EntityManagerFactoryDependsOnPostProcessor subclass as a convenient way of setting up the necessary dependencies. For example, if you use Hibernate Search with Elasticsearch as its index manager, any EntityManagerFactory beans must be configured to depend on the elasticsearchClient bean, as shown in the following example:

If you need to use jOOQ with multiple data sources, you should create your own DSLContext for each one. Refer to JooqAutoConfiguration for more details.

JPA has features for DDL generation, and these can be set up to run on startup against the database. This is controlled through two external properties:

You can set spring.jpa.hibernate.ddl-auto explicitly and the standard Hibernate property values are none, validate, update, create, and create-drop. Spring Boot chooses a default value for you based on whether it thinks your database is embedded. It defaults to create-drop if no schema manager has been detected or none in all other cases. An embedded database is detected by looking at the Connection type. hsqldb, h2, and derby are embedded, and others are not. Be careful when switching from in-memory to a ‘real’ database that you do not make assumptions about the existence of the tables and data in the new platform. You either have to set ddl-auto explicitly or use one of the other mechanisms to initialize the database.

In addition, a file named import.sql in the root of the classpath is executed on startup if Hibernate creates the schema from scratch (that is, if the ddl-auto property is set to create or create-drop). This can be useful for demos and for testing if you are careful but is probably not something you want to be on the classpath in production. It is a Hibernate feature (and has nothing to do with Spring).

Spring Boot can automatically create the schema (DDL scripts) of your DataSource and initialize it (DML scripts). It loads SQL from the standard root classpath locations: schema.sql and data.sql, respectively. In addition, Spring Boot processes the schema-${platform}.sql and data-${platform}.sql files (if present), where platform is the value of spring.datasource.platform. This allows you to switch to database-specific scripts if necessary. For example, you might choose to set it to the vendor name of the database (hsqldb, h2, oracle, mysql, postgresql, and so on).

By default, Spring Boot enables the fail-fast feature of the Spring JDBC initializer. This means that, if the scripts cause exceptions, the application fails to start. You can tune that behavior by setting spring.datasource.continue-on-error.

If you are using R2DBC, the regular DataSource auto-configuration backs off so none of the options described above can be used.

If you are using Spring Data R2DBC, you can initialize the database on startup using SQL scripts as shown in the following example:

Alternatively, you can configure either Flyway or Liquibase to configure a DataSource for you for the duration of the migration. Both these libraries offer properties to set the url, username and password of the database to migrate.

If you use Spring Batch, it comes pre-packaged with SQL initialization scripts for most popular database platforms. Spring Boot can detect your database type and execute those scripts on startup. If you use an embedded database, this happens by default. You can also enable it for any database type, as shown in the following example:

You can also switch off the initialization explicitly by setting spring.batch.initialize-schema=never.

Spring Boot supports two higher-level migration tools: Flyway and Liquibase.

If your JMS broker does not support transacted sessions, you have to disable the support of transactions altogether. If you create your own JmsListenerContainerFactory, there is nothing to do, since, by default it cannot be transacted. If you want to use the DefaultJmsListenerContainerFactoryConfigurer to reuse Spring Boot’s default, you can disable transacted sessions, as follows:

The preceding example overrides the default factory, and it should be applied to any other factory that your application defines, if any.

By default, batch applications require a DataSource to store job details. Spring Batch expects a single DataSource by default. To have it use a DataSource other than the application’s main DataSource, declare a DataSource bean, annotating its @Bean method with @BatchDataSource. If you do so and want two data sources, remember to mark the other one @Primary. To take greater control, implement BatchConfigurer. See The Javadoc of @EnableBatchProcessing for more details.

For more info about Spring Batch, see the Spring Batch project page.

Spring Batch auto-configuration is enabled by adding @EnableBatchProcessing to one of your @Configuration classes.

By default, it executes all Jobs in the application context on startup (see JobLauncherApplicationRunner for details). You can narrow down to a specific job or jobs by specifying spring.batch.job.names (which takes a comma-separated list of job name patterns).

See BatchAutoConfiguration and @EnableBatchProcessing for more details.

Spring Boot converts any command line argument starting with -- to a property to add to the Environment, see accessing command line properties. This should not be used to pass arguments to batch jobs. To specify batch arguments on the command line, use the regular format (i.e. without --), as shown in the following example:

If you specify a property of the Environment on the command line, it is ignored by the job. Consider the following command:

This provides only one argument to the batch job: someParameter=someValue.

Spring Batch requires a data store for the Job repository. If you use Spring Boot, you must use an actual database. Note that it can be an in-memory database, see Configuring a Job Repository.

In a standalone application, the Actuator HTTP port defaults to the same as the main HTTP port. To make the application listen on a different port, set the external property: management.server.port. To listen on a completely different network address (such as when you have an internal network for management and an external one for user applications), you can also set management.server.address to a valid IP address to which the server is able to bind.

For more detail, see the ManagementServerProperties source code and “production-ready-features.html” in the “Production-ready features” section.

Spring Boot installs a ‘whitelabel’ error page that you see in a browser client if you encounter a server error (machine clients consuming JSON and other media types should see a sensible response with the right error code).

Overriding the error page with your own depends on the templating technology that you use. For example, if you use Thymeleaf, you can add an error.html template. If you use FreeMarker, you can add an error.ftlh template. In general, you need a View that resolves with a name of error or a @Controller that handles the /error path. Unless you replaced some of the default configuration, you should find a BeanNameViewResolver in your ApplicationContext, so a @Bean named error would be one way of doing that. See ErrorMvcAutoConfiguration for more options.

See also the section on “Error Handling” for details of how to register handlers in the servlet container.

Information returned by the env and configprops endpoints can be somewhat sensitive so keys matching a certain pattern are sanitized by default (i.e. their values are replaced by ******).

The patterns to use can be customized using the management.endpoint.env.keys-to-sanitize and management.endpoint.configprops.keys-to-sanitize respectively.

Spring Boot uses sensible defaults for such keys: any key ending with the word "password", "secret", "key", "token", "vcap_services", "sun.java.command" is entirely sanitized. Additionally, any key that holds the word credentials as part of the key is sanitized (configured as a regular expression, i.e. *credentials.*).

Furthermore, Spring Boot only sanitizes the sensitive portion of URIs for keys which end with "uri", "uris", "address", or "addresses". The sensitive portion of the URI is identified using the format <scheme>://<username>:<password>@<host>:<port>/. For example, for the property myclient.uri=http://user1:password1@localhost:8081, the resulting sanitized value is http://user1:******@localhost:8081.

Spring Boot health indicators return a Status type to indicate the overall system health. If you want to monitor or alert on levels of health for a particular application, you can export these statuses as metrics via Micrometer. By default, the status codes “UP”, “DOWN”, “OUT_OF_SERVICE” and “UNKNOWN” are used by Spring Boot. To export these, you’ll need to convert these states to some set of numbers so that they can be used with a Micrometer Gauge.

The following example shows one way to write such an exporter:

If you define a @Configuration with a WebSecurityConfigurerAdapter in your application, it switches off the default webapp security settings in Spring Boot.

If you provide a @Bean of type AuthenticationManager, AuthenticationProvider, or UserDetailsService, the default @Bean for InMemoryUserDetailsManager is not created. This means you have the full feature set of Spring Security available (such as various authentication options).

The easiest way to add user accounts is to provide your own UserDetailsService bean.

Ensuring that all your main endpoints are only available over HTTPS is an important chore for any application. If you use Tomcat as a servlet container, then Spring Boot adds Tomcat’s own RemoteIpValve automatically if it detects some environment settings, and you should be able to rely on the HttpServletRequest to report whether it is secure or not (even downstream of a proxy server that handles the real SSL termination). The standard behavior is determined by the presence or absence of certain request headers (x-forwarded-for and x-forwarded-proto), whose names are conventional, so it should work with most front-end proxies. You can switch on the valve by adding some entries to application.properties, as shown in the following example:

(The presence of either of those properties switches on the valve. Alternatively, you can add the RemoteIpValve by customizing the TomcatServletWebServerFactory using a WebServerFactoryCustomizer bean.)

To configure Spring Security to require a secure channel for all (or some) requests, consider adding your own WebSecurityConfigurerAdapter that adds the following HttpSecurity configuration:

There are several options for hot reloading. The recommended approach is to use spring-boot-devtools, as it provides additional development-time features, such as support for fast application restarts and LiveReload as well as sensible development-time configuration (such as template caching). Devtools works by monitoring the classpath for changes. This means that static resource changes must be "built" for the change to take effect. By default, this happens automatically in Eclipse when you save your changes. In IntelliJ IDEA, the Make Project command triggers the necessary build. Due to the default restart exclusions, changes to static resources do not trigger a restart of your application. They do, however, trigger a live reload.

Alternatively, running in an IDE (especially with debugging on) is a good way to do development (all modern IDEs allow reloading of static resources and usually also allow hot-swapping of Java class changes).

Finally, the Maven and Gradle plugins can be configured (see the addResources property) to support running from the command line with reloading of static files directly from source. You can use that with an external css/js compiler process if you are writing that code with higher-level tools.

Most of the templating technologies supported by Spring Boot include a configuration option to disable caching (described later in this document). If you use the spring-boot-devtools module, these properties are automatically configured for you at development time.

The spring-boot-devtools module includes support for automatic application restarts. While not as fast as technologies such as JRebel it is usually significantly faster than a “cold start”. You should probably give it a try before investigating some of the more complex reload options discussed later in this document.

For more details, see the using-spring-boot.html section.

Many modern IDEs (Eclipse, IDEA, and others) support hot swapping of bytecode. Consequently, if you make a change that does not affect class or method signatures, it should reload cleanly with no side effects.

Both the Maven plugin and the Gradle plugin allow generating build information containing the coordinates, name, and version of the project. The plugins can also be configured to add additional properties through configuration. When such a file is present, Spring Boot auto-configures a BuildProperties bean.

To generate build information with Maven, add an execution for the build-info goal, as shown in the following example:

The following example does the same with Gradle:

Both Maven and Gradle allow generating a git.properties file containing information about the state of your git source code repository when the project was built.

For Maven users, the spring-boot-starter-parent POM includes a pre-configured plugin to generate a git.properties file. To use it, add the following declaration to your POM:

Gradle users can achieve the same result by using the gradle-git-properties plugin, as shown in the following example:

Both the Maven and Gradle plugins allow the properties that are included in git.properties to be configured.

The spring-boot-dependencies POM manages the versions of common dependencies. The Spring Boot plugins for Maven and Gradle allow these managed dependency versions to be customized using build properties.

To override dependency versions with Maven, see this section of the Maven plugin’s documentation.

To override dependency versions in Gradle, see this section of the Gradle plugin’s documentation.

The spring-boot-maven-plugin can be used to create an executable “fat” JAR. If you use the spring-boot-starter-parent POM, you can declare the plugin and your jars are repackaged as follows:

If you do not use the parent POM, you can still use the plugin. However, you must additionally add an <executions> section, as follows:

See the plugin documentation for full usage details.

Like a war file, a Spring Boot application is not intended to be used as a dependency. If your application contains classes that you want to share with other projects, the recommended approach is to move that code into a separate module. The separate module can then be depended upon by your application and other projects.

If you cannot rearrange your code as recommended above, Spring Boot’s Maven and Gradle plugins must be configured to produce a separate artifact that is suitable for use as a dependency. The executable archive cannot be used as a dependency as the executable jar format packages application classes in BOOT-INF/classes. This means that they cannot be found when the executable jar is used as a dependency.

To produce the two artifacts, one that can be used as a dependency and one that is executable, a classifier must be specified. This classifier is applied to the name of the executable archive, leaving the default archive for use as a dependency.

To configure a classifier of exec in Maven, you can use the following configuration:

Most nested libraries in an executable jar do not need to be unpacked in order to run. However, certain libraries can have problems. For example, JRuby includes its own nested jar support, which assumes that the jruby-complete.jar is always directly available as a file in its own right.

To deal with any problematic libraries, you can flag that specific nested jars should be automatically unpacked when the executable jar first runs. Such nested jars are written beneath the temporary directory identified by the java.io.tmpdir system property.

For example, to indicate that JRuby should be flagged for unpacking by using the Maven Plugin, you would add the following configuration:

Often, if you have an executable and a non-executable jar as two separate build products, the executable version has additional configuration files that are not needed in a library jar. For example, the application.yml configuration file might be excluded from the non-executable JAR.

In Maven, the executable jar must be the main artifact and you can add a classified jar for the library, as follows:

To attach a remote debugger to a Spring Boot application that was started with Maven, you can use the jvmArguments property of the maven plugin.

See this example for more details.

To build with Ant, you need to grab dependencies, compile, and then create a jar or war archive. To make it executable, you can either use the spring-boot-antlib module or you can follow these instructions:

The following example shows how to build an executable archive with Ant:

The first step in producing a deployable war file is to provide a SpringBootServletInitializer subclass and override its configure method. Doing so makes use of Spring Framework’s Servlet 3.0 support and lets you configure your application when it is launched by the servlet container. Typically, you should update your application’s main class to extend SpringBootServletInitializer, as shown in the following example:

The next step is to update your build configuration such that your project produces a war file rather than a jar file. If you use Maven and spring-boot-starter-parent (which configures Maven’s war plugin for you), all you need to do is to modify pom.xml to change the packaging to war, as follows:

If you use Gradle, you need to modify build.gradle to apply the war plugin to the project, as follows:

The final step in the process is to ensure that the embedded servlet container does not interfere with the servlet container to which the war file is deployed. To do so, you need to mark the embedded servlet container dependency as being provided.

If you use Maven, the following example marks the servlet container (Tomcat, in this case) as being provided:

If you use Gradle, the following example marks the servlet container (Tomcat, in this case) as being provided:

If you use the Spring Boot build tools, marking the embedded servlet container dependency as provided produces an executable war file with the provided dependencies packaged in a lib-provided directory. This means that, in addition to being deployable to a servlet container, you can also run your application by using java -jar on the command line.

To convert an existing non-web Spring application to a Spring Boot application, replace the code that creates your ApplicationContext and replace it with calls to SpringApplication or SpringApplicationBuilder. Spring MVC web applications are generally amenable to first creating a deployable war application and then migrating it later to an executable war or jar. See the Getting Started Guide on Converting a jar to a war.

To create a deployable war by extending SpringBootServletInitializer (for example, in a class called Application) and adding the Spring Boot @SpringBootApplication annotation, use code similar to that shown in the following example:

Remember that, whatever you put in the sources is merely a Spring ApplicationContext. Normally, anything that already works should work here. There might be some beans you can remove later and let Spring Boot provide its own defaults for them, but it should be possible to get something working before you need to do that.

Static resources can be moved to /public (or /static or /resources or /META-INF/resources) in the classpath root. The same applies to messages.properties (which Spring Boot automatically detects in the root of the classpath).

Vanilla usage of Spring DispatcherServlet and Spring Security should require no further changes. If you have other features in your application (for instance, using other servlets or filters), you may need to add some configuration to your Application context, by replacing those elements from the web.xml, as follows:

Once the war file is working, you can make it executable by adding a main method to your Application, as shown in the following example:

Applications can fall into more than one category:

All of these should be amenable to translation, but each might require slightly different techniques.

Servlet 3.0+ applications might translate pretty easily if they already use the Spring Servlet 3.0+ initializer support classes. Normally, all the code from an existing WebApplicationInitializer can be moved into a SpringBootServletInitializer. If your existing application has more than one ApplicationContext (for example, if it uses AbstractDispatcherServletInitializer) then you might be able to combine all your context sources into a single SpringApplication. The main complication you might encounter is if combining does not work and you need to maintain the context hierarchy. See the entry on building a hierarchy for examples. An existing parent context that contains web-specific features usually needs to be broken up so that all the ServletContextAware components are in the child context.

Applications that are not already Spring applications might be convertible to Spring Boot applications, and the previously mentioned guidance may help. However, you may yet encounter problems. In that case, we suggest asking questions on Stack Overflow with a tag of spring-boot.

To deploy a Spring Boot application to WebLogic, you must ensure that your servlet initializer directly implements WebApplicationInitializer (even if you extend from a base class that already implements it).

A typical initializer for WebLogic should resemble the following example:

If you use Logback, you also need to tell WebLogic to prefer the packaged version rather than the version that was pre-installed with the server. You can do so by adding a WEB-INF/weblogic.xml file with the following contents:

By default, the Spring Boot starter (spring-boot-starter-data-redis) uses Lettuce. You need to exclude that dependency and include the Jedis one instead. Spring Boot manages both of these dependencies so you can switch to Jedis without specifying a version.

The following example shows how to do so in Maven:

The following example shows how to do so in Gradle:

The Testcontainers library provides a way to manage services running inside Docker containers. It integrates with JUnit, allowing you to write a test class that can start up a container before any of the tests run. Testcontainers is especially useful for writing integration tests that talk to a real backend service such as MySQL, MongoDB, Cassandra etc. Testcontainers can be used in a Spring Boot test as follows:

This will start up a docker container running Neo4j (if Docker is running locally) before any of the tests are run. In most cases, you will need to configure the application using details from the running container, such as container IP or port.

This can be done with a static @DynamicPropertySource method that allows adding dynamic property values to the Spring Environment.

The above configuration allows Neo4j-related beans in the application to communicate with Neo4j running inside the Testcontainers-managed Docker container.