Home Original page

Spring Boot 3.0 Release Notes

Java 17 Baseline and Java 19 Support

Spring Boot 3.0 requires Java 17 as a minimum version. If you are currently using Java 8 or Java 11, you’ll need to upgrade your JDK before you can develop Spring Boot 3.0 applications.

Spring Boot 3.0 also works well, and has been tested with JDK 19.

GraalVM Baseline and Native Build Tools

Spring Boot requires Graal 22.3 or later and Native Build Tools Plugin 0.9.17 or later

Third-party Library Upgrades

Other Spring projects upgraded in this release include:

Spring Boot 3.0 has migrated from Java EE to Jakarta EE APIs for all dependencies. Wherever possible, Jakarta EE 10 compatible dependencies have been chosen, including:

  • Jakarta Activation 2.1

  • Jakarta JMS 3.1

  • Jakarta JSON 2.1

  • Jakarta JSON Bind 3.0

  • Jakarta Mail 2.1

  • Jakarta Persistence 3.1

  • Jakarta Servlet 6.0

  • Jakarta Servlet JSP JSTL 3.0

  • Jakarta Transaction 2.0

  • Jakarta Validation 3.0

  • Jakarta WebSocket 2.1

  • Jakarta WS RS 3.1

  • Jakarta XML SOAP 3.0

  • Jakarta XML WS 4.0

We’ve also upgraded to the latest stable releases of third-party jars wherever possible. Some notable dependency upgrades here include:

GraalVM Native Image Support

Spring Boot 3.0 applications can now be converted into GraalVM native images which can provide significant memory and startup-up performance improvements. Supporting GraalVM Native Images has been a major engineering effort undertaken across the entire Spring portfolio.

Log4j2 Enhancements

Log4j2 support has been updated with new extensions that provide the following functionality:

  • Profile-specific Configuration

  • Environment Properties Lookup

  • Log4j2 System Properties

Improved @ConstructorBinding Detection

When using constructor bound @ConfigurationProperties the @ConstructorBinding annotation is no longer required if the class has a single parameterized constructor. If you have more than one constructor, you’ll still need to use @ConstructorBinding to tell Spring Boot which one to use.

For most users, this updated logic will allow for simpler @ConfigurationProperties classes. If, however, you have a @ConfigurationProperties and you want to inject beans into the constructor rather than binding it, you’ll now need to add an @Autowired annotation.

Micrometer Updates

Auto-configuration for Micrometer Observation API

Spring Boot 3.0 supports the new observation APIs introduced in Micrometer 1.10. The new ObservationRegistry interface can be used to create observations which provides a single API for both metrics and traces. Spring Boot now auto-configures an instance of ObservationRegistry for you.

If micrometer-core is on the classpath, a DefaultMeterObservationHandler is registered on the ObservationRegistry, which means that every stopped Observation leads to a timer. ObservationPredicate, GlobalObservationConvention and ObservationHandler are automatically registered on the ObservationRegistry. You can use ObservationRegistryCustomizer to further customize the ObservationRegistry if you need to.

Auto-configuration for Micrometer Tracing

Spring Boot now auto-configures Micrometer Tracing for you. This includes support for Brave, OpenTelemetry, Zipkin and Wavefront.

When using the Micrometer Observation API, finishing observations will lead to spans reported to Zipkin or Wavefront. Tracing can be controlled with properties under management.tracing. Zipkin can be configured with management.zipkin.tracing, while Wavefront uses management.wavefront.

More details, including the various dependencies that you’ll need to add, are in the tracing section of the reference documentation.

Auto-configuration for Micrometer’s OtlpMeterRegistry

An OtlpMeterRegistry is now auto-configured when io.micrometer:micrometer-registry-otlp is on the classpath. The meter registry can be configured using management.otlp.metrics.export.* properties.

Prometheus Support

Auto-Configuration for Prometheus Exemplars

When there is a Micrometer Tracing Tracer bean and Prometheus is on the classpath, a SpanContextSupplier is now auto-configured. This supplier links metrics to traces by making the current trace ID and span ID available to Prometheus.

Making a PUT to Prometheus Push Gateway on Shutdown

The Push Gateway can be configured to perform a PUT on shutdown. To do so, set management.prometheus.metrics.export.pushgateway.shutdown-operation to put. Additionally, the existing push setting has been deprecated and post should now be used instead.

More Flexible Auto-configuration for Spring Data JDBC

The auto-configuration for Spring Data JDBC is now more flexible. Several auto-configured beans that are required by Spring Data JDBC are now conditional and can be replaced by defining a bean of the same type. The types of the beans that can now be replaced are the following:

  • org.springframework.data.jdbc.core.JdbcAggregateTemplate

  • org.springframework.data.jdbc.core.convert.DataAccessStrategy

  • org.springframework.data.jdbc.core.convert.JdbcConverter

  • org.springframework.data.jdbc.core.convert.JdbcCustomConversions

  • org.springframework.data.jdbc.core.mapping.JdbcMappingContext

  • org.springframework.data.relational.RelationalManagedTypes

  • org.springframework.data.relational.core.dialect.Dialect

Enabling Async Acks with Apache Kafka

A new configuration property, spring.kafka.listener.async-acks, for enabling async acks with Kafka has been added. To enable async acks, set the property to true. The property only applies when spring.kafka.listener.async-mode is set to manual or manual-immediate.

Elasticsearch Java Client

Auto-configuration for the new Elasticsearch Java Client has been introduced. It can be configured using the existing spring.elasticsearch.* configuration properties.

The auto-configuration for the new client does not use the auto-configured ObjectMapper for JSON mapping. This is to prevent clashes between the needs of the application and the needs of Elasticsearch. To control the ObjectMapper that is used, define a JacksonJsonpMapper bean. For example, the following configuration will result in the Elastic client using the context’s ObjectMapper:

@Bean
JacksonJsonpMapper jacksonJsonpMapper(ObjectMapper objectMapper) {
	return new JacksonJsonpMapper(objectMapper);
}

Auto-configuration of JdkClientHttpConnector

In the absence of Reactor Netty, Jetty’s reactive client, and the Apache HTTP client a JdkClientHttpConnector will now be auto-configured. This allows WebClient to be used with the JDK’s HttpClient.

@SpringBootTest with Main Methods

The @SpringBootTest annotation can now use the main of any discovered @SpringBootConfiguration class if it’s available. This means that any custom SpringApplication configuration performed by your main method can now be picked up by tests.

To use the main method for a test set the useMainMethod attribute of @SpringBootTest to UseMainMethod.ALWAYS or UseMainMethod.WHEN_AVAILABLE.

Miscellaneous

Apart from the changes listed above, there have also been some minor tweaks and improvements including:

  • Host names are no longer logged during application startup. This prevents a network lookup which helps improve startup times.

  • Support for Java’s SecurityManager has been removed following its deprecation in the JDK.

  • Support for Spring Framework’s CommonsMultipartResolver has been removed following its removal in Spring Framework 6.

  • The spring.mvc.ignore-default-model-on-redirect has been deprecated to align with upstream Spring Framework changes.

  • WebJars resource handler path pattern can be customized using spring.mvc.webjars-path-pattern or spring.webflux.webjars-path-pattern.

  • Trusted proxies of Tomcat’s remote IP valve can be configured using server.tomcat.remoteip.trusted-proxies.

  • Bean Validation Configuration can now be customised by defining a ValidationConfigurationCustomizer bean.

  • Log4j2’s Log4jBridgeHandler is now used to route JUL-based logging into Log4j2 rather than routing through SLF4J.

  • Bean implementing the MeterBinder interface are now bound to meter registries only after all singleton beans have been initialized.

  • SpanCustomizer beans for Brave and OpenTelemetry are now auto-configured.

  • Micrometer’s JvmCompilationMetrics are now auto-configured.

  • DiskSpaceHealthIndicator now includes its path in its log message and its health details.

  • A DataSourceBuilder can now be derived from a wrapped DataSource.

  • Multiple hosts can now be configured for MongoDB using the spring.data.mongodb.additional-hosts property.

  • Elasticsearch’s socketKeepAlive property can be configured using the spring.elasticsearch.socket-keep-alive property.

  • When using spring-rabbit-stream, a RabbitStreamTemplate and Environment will now be auto-configured whether or not spring.rabbitmq.listener.type is stream.

  • Existing Kafka topics can be modified using spring.kafka.admin.modify-topic-configs.

  • WebDriverScope and WebDriverTestExecutionListener have been made public to ease the use of WebDriver in custom test setup.