Development-time Services
Development-time services provide external dependencies needed to run the application while developing it. They are only supposed to be used while developing and are disabled when the application is deployed.
Spring Boot offers support for two development time services, Docker Compose and Testcontainers. The next sections will provide more details about them.
Docker Compose Support
Docker Compose is a popular technology that can be used to define and manage multiple containers for services that your application needs.
A compose.yml
file is typically created next to your application which defines and configures service containers.
A typical workflow with Docker Compose is to run docker compose up
, work on your application with it connecting to started services, then run docker compose down
when you are finished.
The spring-boot-docker-compose
module can be included in a project to provide support for working with containers using Docker Compose.
Add the module dependency to your build, as shown in the following listings for Maven and Gradle:
<dependencies>
<dependency>
<groupId>org.springframework.boot</groupId>
<artifactId>spring-boot-docker-compose</artifactId>
<optional>true</optional>
</dependency>
</dependencies>
dependencies {
developmentOnly("org.springframework.boot:spring-boot-docker-compose")
}
When this module is included as a dependency Spring Boot will do the following:
-
Search for a
compose.yml
and other common compose filenames in your working directory -
Call
docker compose up
with the discoveredcompose.yml
-
Create service connection beans for each supported container
-
Call
docker compose stop
when the application is shutdown
If the Docker Compose services are already running when starting the application, Spring Boot will only create the service connection beans for each supported container.
It will not call docker compose up
again and it will not call docker compose stop
when the application is shutdown.
Repackaged archives do not contain Spring Boot’s Docker Compose by default.
If you want to use this support, you need to include it.
When using the Maven plugin, set the excludeDockerCompose property to false .
When using the Gradle plugin, configure the task’s classpath to include the developmentOnly configuration.
|
Prerequisites
You need to have the docker
and docker compose
(or docker-compose
) CLI applications on your path.
The minimum supported Docker Compose version is 2.2.0.
Service Connections
A service connection is a connection to any remote service. Spring Boot’s auto-configuration can consume the details of a service connection and use them to establish a connection to a remote service. When doing so, the connection details take precedence over any connection-related configuration properties.
When using Spring Boot’s Docker Compose support, service connections are established to the port mapped by the container.
Docker compose is usually used in such a way that the ports inside the container are mapped to ephemeral ports on your computer. For example, a Postgres server may run inside the container using port 5432 but be mapped to a totally different port locally. The service connection will always discover and use the locally mapped port. |
Service connections are established by using the image name of the container. The following service connections are currently supported:
Connection Details | Matched on |
---|---|
|
Containers named "symptoma/activemq" or "apache/activemq-classic" |
|
Containers named "apache/activemq-artemis" |
|
Containers named "cassandra" or "bitnami/cassandra" |
|
Containers named "elasticsearch" or "bitnami/elasticsearch" |
|
Containers named "gvenzl/oracle-free", "gvenzl/oracle-xe", "mariadb", "bitnami/mariadb", "mssql/server", "mysql", "bitnami/mysql", "postgres", or "bitnami/postgresql" |
|
Containers named "osixia/openldap" |
|
Containers named "mongo" or "bitnami/mongodb" |
|
Containers named "neo4j" or "bitnami/neo4j" |
|
Containers named "otel/opentelemetry-collector-contrib" |
|
Containers named "otel/opentelemetry-collector-contrib" |
|
Containers named "apachepulsar/pulsar" |
|
Containers named "gvenzl/oracle-free", "gvenzl/oracle-xe", "mariadb", "bitnami/mariadb", "mssql/server", "mysql", "bitnami/mysql", "postgres", or "bitnami/postgresql" |
|
Containers named "rabbitmq" or "bitnami/rabbitmq" |
|
Containers named "redis" or "bitnami/redis" |
|
Containers named "openzipkin/zipkin". |
Custom Images
Sometimes you may need to use your own version of an image to provide a service. You can use any custom image as long as it behaves in the same way as the standard image. Specifically, any environment variables that the standard image supports must also be used in your custom image.
If your image uses a different name, you can use a label in your compose.yml
file so that Spring Boot can provide a service connection.
Use a label named org.springframework.boot.service-connection
to provide the service name.
For example:
services:
redis:
image: 'mycompany/mycustomredis:7.0'
ports:
- '6379'
labels:
org.springframework.boot.service-connection: redis
Skipping Specific Containers
If you have a container image defined in your compose.yml
that you don’t want connected to your application you can use a label to ignore it.
Any container with labeled with org.springframework.boot.ignore
will be ignored by Spring Boot.
For example:
services:
redis:
image: 'redis:7.0'
ports:
- '6379'
labels:
org.springframework.boot.ignore: true
Using a Specific Compose File
If your compose file is not in the same directory as your application, or if it’s named differently, you can use spring.docker.compose.file
in your application.properties
or application.yaml
to point to a different file.
Properties can be defined as an exact path or a path that’s relative to your application.
For example:
-
Properties
-
YAML
spring.docker.compose.file=../my-compose.yml
spring:
docker:
compose:
file: "../my-compose.yml"
Waiting for Container Readiness
Containers started by Docker Compose may take some time to become fully ready.
The recommended way of checking for readiness is to add a healthcheck
section under the service definition in your compose.yml
file.
Since it’s not uncommon for healthcheck
configuration to be omitted from compose.yml
files, Spring Boot also checks directly for service readiness.
By default, a container is considered ready when a TCP/IP connection can be established to its mapped port.
You can disable this on a per-container basis by adding a org.springframework.boot.readiness-check.tcp.disable
label in your compose.yml
file.
For example:
services:
redis:
image: 'redis:7.0'
ports:
- '6379'
labels:
org.springframework.boot.readiness-check.tcp.disable: true
You can also change timeout values in your application.properties
or application.yaml
file:
-
Properties
-
YAML
spring.docker.compose.readiness.tcp.connect-timeout=10s
spring.docker.compose.readiness.tcp.read-timeout=5s
spring:
docker:
compose:
readiness:
tcp:
connect-timeout: 10s
read-timeout: 5s
The overall timeout can be configured using spring.docker.compose.readiness.timeout
.
Controlling the Docker Compose Lifecycle
By default Spring Boot calls docker compose up
when your application starts and docker compose stop
when it’s shut down.
If you prefer to have different lifecycle management you can use the spring.docker.compose.lifecycle-management
property.
The following values are supported:
-
none
- Do not start or stop Docker Compose -
start-only
- Start Docker Compose when the application starts and leave it running -
start-and-stop
- Start Docker Compose when the application starts and stop it when the JVM exits
In addition you can use the spring.docker.compose.start.command
property to change whether docker compose up
or docker compose start
is used.
The spring.docker.compose.stop.command
allows you to configure if docker compose down
or docker compose stop
is used.
The following example shows how lifecycle management can be configured:
-
Properties
-
YAML
spring.docker.compose.lifecycle-management=start-and-stop
spring.docker.compose.start.command=start
spring.docker.compose.stop.command=down
spring.docker.compose.stop.timeout=1m
spring:
docker:
compose:
lifecycle-management: start-and-stop
start:
command: start
stop:
command: down
timeout: 1m
Activating Docker Compose Profiles
Docker Compose profiles are similar to Spring profiles in that they let you adjust your Docker Compose configuration for specific environments.
If you want to activate a specific Docker Compose profile you can use the spring.docker.compose.profiles.active
property in your application.properties
or application.yaml
file:
-
Properties
-
YAML
spring.docker.compose.profiles.active=myprofile
spring:
docker:
compose:
profiles:
active: "myprofile"
Using Docker Compose in Tests
By default, Spring Boot’s Docker Compose support is disabled when running tests.
To enable Docker Compose support in tests, set spring.docker.compose.skip.in-tests
to false
.
When using Gradle, you also need to change the configuration of the spring-boot-docker-compose
dependency from developmentOnly
to testAndDevelopmentOnly
:
dependencies {
testAndDevelopmentOnly("org.springframework.boot:spring-boot-docker-compose")
}
Testcontainers Support
As well as using Testcontainers for integration testing, it’s also possible to use them at development time. The next sections will provide more details about that.
Using Testcontainers at Development Time
This approach allows developers to quickly start containers for the services that the application depends on, removing the need to manually provision things like database servers. Using Testcontainers in this way provides functionality similar to Docker Compose, except that your container configuration is in Java rather than YAML.
To use Testcontainers at development time you need to launch your application using your “test” classpath rather than “main”. This will allow you to access all declared test dependencies and give you a natural place to write your test configuration.
To create a test launchable version of your application you should create an “Application” class in the src/test
directory.
For example, if your main application is in src/main/java/com/example/MyApplication.java
, you should create src/test/java/com/example/TestMyApplication.java
The TestMyApplication
class can use the SpringApplication.from(…)
method to launch the real application:
import org.springframework.boot.SpringApplication;
public class TestMyApplication {
public static void main(String[] args) {
SpringApplication.from(MyApplication::main).run(args);
}
}
You’ll also need to define the Container
instances that you want to start along with your application.
To do this, you need to make sure that the spring-boot-testcontainers
module has been added as a test
dependency.
Once that has been done, you can create a @TestConfiguration
class that declares @Bean
methods for the containers you want to start.
You can also annotate your @Bean
methods with @ServiceConnection
in order to create ConnectionDetails
beans.
See the service connections section for details of the supported technologies.
A typical Testcontainers configuration would look like this:
import org.testcontainers.containers.Neo4jContainer;
import org.springframework.boot.test.context.TestConfiguration;
import org.springframework.boot.testcontainers.service.connection.ServiceConnection;
import org.springframework.context.annotation.Bean;
@TestConfiguration(proxyBeanMethods = false)
public class MyContainersConfiguration {
@Bean
@ServiceConnection
public Neo4jContainer<?> neo4jContainer() {
return new Neo4jContainer<>("neo4j:5");
}
}
The lifecycle of Container beans is automatically managed by Spring Boot.
Containers will be started and stopped automatically.
|
You can use the spring.testcontainers.beans.startup property to change how containers are started.
By default sequential startup is used, but you may also choose parallel if you wish to start multiple containers in parallel.
|
Once you have defined your test configuration, you can use the with(…)
method to attach it to your test launcher:
import org.springframework.boot.SpringApplication;
public class TestMyApplication {
public static void main(String[] args) {
SpringApplication.from(MyApplication::main).with(MyContainersConfiguration.class).run(args);
}
}
You can now launch TestMyApplication
as you would any regular Java main
method application to start your application and the containers that it needs to run.
You can use the Maven goal spring-boot:test-run or the Gradle task bootTestRun to do this from the command line.
|
Contributing Dynamic Properties at Development Time
If you want to contribute dynamic properties at development time from your Container
@Bean
methods, you can do so by injecting a DynamicPropertyRegistry
.
This works in a similar way to the @DynamicPropertySource
annotation that you can use in your tests.
It allows you to add properties that will become available once your container has started.
A typical configuration would look like this:
import org.testcontainers.containers.MongoDBContainer;
import org.springframework.boot.test.context.TestConfiguration;
import org.springframework.context.annotation.Bean;
import org.springframework.test.context.DynamicPropertyRegistry;
@TestConfiguration(proxyBeanMethods = false)
public class MyContainersConfiguration {
@Bean
public MongoDBContainer mongoDbContainer(DynamicPropertyRegistry properties) {
MongoDBContainer container = new MongoDBContainer("mongo:5.0");
properties.add("spring.data.mongodb.host", container::getHost);
properties.add("spring.data.mongodb.port", container::getFirstMappedPort);
return container;
}
}
Using a @ServiceConnection is recommended whenever possible, however, dynamic properties can be a useful fallback for technologies that don’t yet have @ServiceConnection support.
|
Importing Testcontainer Declaration Classes
A common pattern when using Testcontainers is to declare Container
instances as static fields.
Often these fields are defined directly on the test class.
They can also be declared on a parent class or on an interface that the test implements.
For example, the following MyContainers
interface declares mongo
and neo4j
containers:
import org.testcontainers.containers.MongoDBContainer;
import org.testcontainers.containers.Neo4jContainer;
import org.testcontainers.junit.jupiter.Container;
import org.springframework.boot.testcontainers.service.connection.ServiceConnection;
public interface MyContainers {
@Container
@ServiceConnection
MongoDBContainer mongoContainer = new MongoDBContainer("mongo:5.0");
@Container
@ServiceConnection
Neo4jContainer<?> neo4jContainer = new Neo4jContainer<>("neo4j:5");
}
If you already have containers defined in this way, or you just prefer this style, you can import these declaration classes rather than defining your containers as @Bean
methods.
To do so, add the @ImportTestcontainers
annotation to your test configuration class:
import org.springframework.boot.test.context.TestConfiguration;
import org.springframework.boot.testcontainers.context.ImportTestcontainers;
@TestConfiguration(proxyBeanMethods = false)
@ImportTestcontainers(MyContainers.class)
public class MyContainersConfiguration {
}
If you don’t intend to use the service connections feature but want to use @DynamicPropertySource instead, remove the @ServiceConnection annotation from the Container fields.
You can also add @DynamicPropertySource annotated methods to your declaration class.
|
Using DevTools with Testcontainers at Development Time
When using devtools, you can annotate beans and bean methods with @RestartScope
.
Such beans won’t be recreated when the devtools restart the application.
This is especially useful for Testcontainer Container
beans, as they keep their state despite the application restart.
import org.testcontainers.containers.MongoDBContainer;
import org.springframework.boot.devtools.restart.RestartScope;
import org.springframework.boot.test.context.TestConfiguration;
import org.springframework.boot.testcontainers.service.connection.ServiceConnection;
import org.springframework.context.annotation.Bean;
@TestConfiguration(proxyBeanMethods = false)
public class MyContainersConfiguration {
@Bean
@RestartScope
@ServiceConnection
public MongoDBContainer mongoDbContainer() {
return new MongoDBContainer("mongo:5.0");
}
}
If you’re using Gradle and want to use this feature, you need to change the configuration of the spring-boot-devtools dependency from developmentOnly to testAndDevelopmentOnly .
With the default scope of developmentOnly , the bootTestRun task will not pick up changes in your code, as the devtools are not active.
|