Advanced options
Container labels
To add a custom label to the container, use withLabel
:
public GenericContainer containerWithLabel = new GenericContainer(DockerImageName.parse("alpine:3.17"))
.withLabel("key", "value");
Additionally, multiple labels may be applied together from a map:
private Map<String, String> mapOfLabels = new HashMap<>();
// populate map, e.g. mapOfLabels.put("key1", "value1");
public GenericContainer containerWithMultipleLabels = new GenericContainer(DockerImageName.parse("alpine:3.17"))
.withLabels(mapOfLabels);
Image Pull Policy
By default, the container image is retrieved from the local Docker images cache. This works well when running against a specific version, but for images with a static tag (i.e. 'latest') this may lead to a newer version not being pulled.
It is possible to specify an Image Pull Policy to determine at runtime whether an image should be pulled or not:
GenericContainer<?> container = new GenericContainer<>(imageName)
.withImagePullPolicy(PullPolicy.alwaysPull())
... or providing a function:
GenericContainer<?> container = new GenericContainer<>(imageName)
.withImagePullPolicy(
new AbstractImagePullPolicy() {
@Override
protected boolean shouldPullCached(DockerImageName imageName, ImageData localImageData) {
return System.getenv("ALWAYS_PULL_IMAGE") != null;
}
}
)
You can also configure Testcontainers to use your custom implementation by using pull.policy
pull.policy=com.mycompany.testcontainers.ExampleImagePullPolicy
Please see the documentation on configuration mechanisms for more information.
Customizing the container
Using docker-java
It is possible to use the docker-java
API directly to customize containers before creation. This is useful if there is a need to use advanced Docker features that are not exposed by the Testcontainers API. Any customizations you make using withCreateContainerCmdModifier
will be applied on top of the container definition that Testcontainers creates, but before it is created.
For example, this can be used to change the container hostname:
@Rule
public GenericContainer theCache = new GenericContainer<>(DockerImageName.parse("redis:6-alpine"))
.withCreateContainerCmdModifier(cmd -> cmd.withHostName("the-cache"));
... or modify container memory (see this if it does not appear to work):
private long memoryInBytes = 32l * 1024l * 1024l;
private long memorySwapInBytes = 64l * 1024l * 1024l;
@Rule
public GenericContainer memoryLimitedRedis = new GenericContainer<>(DockerImageName.parse("redis:6-alpine"))
.withCreateContainerCmdModifier(cmd -> {
cmd.getHostConfig()
.withMemory(memoryInBytes)
.withMemorySwap(memorySwapInBytes);
});
Note
It is recommended to use this sparingly, and follow changes to the docker-java
API if you choose to use this.
It is typically quite stable, though.
For what is possible, consult the docker-java CreateContainerCmd
source code.
Using CreateContainerCmdModifier
Testcontainers provides a CreateContainerCmdModifier
to customize docker-java CreateContainerCmd
via Service Provider Interface (SPI) mechanism.
package org.testcontainers.custom;
import com.github.dockerjava.api.command.CreateContainerCmd;
import org.testcontainers.core.CreateContainerCmdModifier;
import java.util.HashMap;
import java.util.Map;
public class TestCreateContainerCmdModifier implements CreateContainerCmdModifier {
@Override
public CreateContainerCmd modify(CreateContainerCmd createContainerCmd) {
Map<String, String> labels = new HashMap<>();
labels.put("project", "testcontainers-java");
labels.put("scope", "global");
createContainerCmd.getLabels().putAll(labels);
return createContainerCmd;
}
}
The previous implementation should be registered in META-INF/services/org.testcontainers.core.CreateContainerCmdModifier
file.
Warning
CreateContainerCmdModifier
implementation will apply to all containers created by Testcontainers.
Parallel Container Startup
Usually, containers are started sequentially when more than one container is used.
Using Startables.deepStart(container1, container2, ...).join()
will start all containers in parallel.
This can be advantageous to reduce the impact of the container startup overhead.