SKM IT World

Just another blog about IT


4 Comments

Test Coverage Reports For Maven Projects In SonarQube 8.3.x

Some years ago I write a blog post about how to generate test reports in SonarQube separate in test report for unit tests and for integration tests. Since SonarQube 6.2 the test report isn’t separate in these categories any more (see SonarQube’s blog post). SonarQube merges all test reports to one test report with an overall coverage. So how to configure JaCoCo Maven Plugin if you have separate your tests in unit tests (running by Maven Surefire Plugin) and integration tests (running by Maven Failsafe Plugin) in your Maven project.

In the following sections, a solution is presented that meets following criteria:

  • Maven is used as build tool.
  • The project can be a multi module project.
  • Unit tests and integration tests are parts of each module.
  • Test coverage is measured by JaCoCo Maven Plugin.

The road map for the next section is that firstly the Maven project structure is shown for the separation of unit and integration tests. Then the Maven project configuration is shown for having separate unit test runs and integration test runs.  After that, we have a look on the Maven project configuration for the test report generation that covers unit and integration tests. At the end, SonarQube’s configuration is shown for the test report visualization in the SonarQube’s dashboard.

Maven Project Structure

At first, we look at how a default Maven project structure looks like for a single module project.

my-app
├── pom.xml
├── src
│   ├── main
│   │   └── java
│   └── test
│       └── java

The directory src/main/java contains the production source code and the directory src/test/java contains the test source code. We could put unit tests and integration tests together in this directory. But we want to separate these two types of tests in separate directories. Therefore, we add a new directory called src/it/java. Then unit tests are put in the directory src/test/java and the integration tests are put in the directory src/it/java, so the new project structure looks like the following one.

my-app
├── pom.xml
├── src
│   ├── it
│   │   └── java
│   ├── main
│   │   └── java
│   └── test
│       └── java

Unit And Integration Test Runs

Fortunately, the unit test run configuration is a part of the Maven default project configuration. Maven runs these tests automatically if following criteria are met:

  • The tests are in the directory src/test/java and
  • the test class name either starts with Test or ends with Test or TestCase.

Maven runs these tests during the Maven’s build lifecylce phase test.

The integration test run configuration has to be done manually. It exists Maven plugins that can help. We want that the following criteria are met:

  • integration tests are stored in the directory src/it/java and
  • the integration test class name either starts IT or ends with IT or ITCase and
  • integrations tests runs during the Maven’s build lifecycle phase integration-test.

Firstly, Maven has to know that it has to include the directory src/it/java to its test class path. Here, the Build Helper Maven Plugin can help. It adds the directory src/it/java to the test class path.

            <plugin>
                <groupId>org.codehaus.mojo</groupId>
                <artifactId>build-helper-maven-plugin</artifactId>
                <version>3.1.0</version>
                <executions>
                    <execution>
                        <goals>
                            <goal>add-test-source</goal>
                            <goal>add-test-resource</goal>
                        </goals>
                        <configuration>
                            <sources>
                                <source>src/it/java</source>
                            </sources>
                            <resources>
                                <resource>
                                    <directory>src/it/resources</directory>
                                </resource>
                            </resources>
                        </configuration>
                    </execution>
                </executions>
            </plugin>

The above code snippet has to be inserted into the section <project><build><plugins> in the project root pom.

Maven’s build lifecycle contains a phase called integration-test.  In this phase, we want to run the integration test. Fortunately, Maven Failsafe Plugins’ goal integration-test is bind to this phase automatically when it’s set up in the POM. If you want that the build fails when the integration tests fails then the goal verify also has to be added into the POM:

            <plugin>
                <groupId>org.apache.maven.plugins</groupId>
                <artifactId>maven-failsafe-plugin</artifactId>
                <version>3.0.0-M4</version>
                <configuration>
                    <encoding>${project.build.sourceEncoding}</encoding>
                </configuration>
                <executions>
                    <execution>
                        <goals>
                            <goal>integration-test</goal>
                            <goal>verify</goal>
                        </goals>
                    </execution>
                </executions>
            </plugin>

Again, the above code snippet also has to be inserted into the section <project><build><plugins> in the project root pom. Then Maven Failsafe Plugin runs the integration tests automatically, when their class name either starts with IT or ends with IT or ITCase.

Test Report Generation

We want to use the JaCoCo Maven Plugin for the test report generation. It should generate test reports for the unit tests and for the integration tests. Therefore, the plugin has to two separated agents, that have to be prepared. Then they generate the report during the test runs. The Maven’s build lifecycle contains own phases for preparation before the test phases (test and integration-test). The preparation phase for the test phase is called process-test-classes and the preparation phase for integration-test phase is called pre-integration-test. JaCoCo binds its agent to these phases automatically, when its goals prepare-agent and prepare-agent-integration are set up in the POM. But this is not enough. JaCoCo also has to create a report, so that SonarQube can read the reports for the visualization. Therefore, we have to add the goals report and report-integration in the POM:

            <plugin>
                <groupId>org.jacoco</groupId>
                <artifactId>jacoco-maven-plugin</artifactId>
                <version>0.8.5</version>
                <executions>    
                    <execution>
                        <goals>  
                            <goal>prepare-agent</goal>
                            <goal>prepare-agent-integration</goal>
                            <goal>report</goal>
                            <goal>report-integration</goal>
                        </goals>  
                    </execution>
                </executions>  
            </plugin>

Again, it is a part of the section <project><build><plugins>.

Now, we can run the goal mvn verify and our project is built inclusive unit and integration test and inclusive generating two test reports.

SonarQube Test Report Visualization

Now, we want to visualize our test reports in SonarQube. Therefore, we have to run the Sonar Maven 3 Plugin (command mvn sonar:sonar)  in our project after a successful build. So Sonar Maven Plugin knows where to upload the report, we have to configure our SonarQube instance in ~/.m2/setting.xml:

    <profile>
      <id>sonar</id>
      <activation>
        <activeByDefault>true</activeByDefault>
      </activation>
      <properties>
        <!-- Optional URL to server. Default value is http://localhost:9000 -->
        <sonar.host.url>http://localhost:9000</sonar.host.url>
      </properties>
    </profile>

When we open our project in the SonarQube dashboard, we see the overall test coverage report.

Summary

This blog describes how to generate test reports for a Maven build if unit tests and integration tests run separately. On GitHub, I host a sample project that demonstrate all configuration steps. As technical environment I use

  • Maven 3.6.3
  • Maven Plugins:
    • Maven Surefire Plugin
    • Maven Failsafe Plugin
    • Build Helper Maven Plugin
    • Jacoco Maven Plugin
    • Sonar Maven Plugin
  • SonarQube 8.3.1
  • Java 11

Links

  1. Jacoco Maven plugin project site
  2. Maven Failsafe Plugin  project site
  3. Build Helper Maven Plugin project site
  4. SonarQube documentation about Test Coverage in common
  5. A sample Maven project on GitHub


Leave a comment

Using Testcontainers in Spring Boot Tests For Database Integration Tests

In this blog post I’d like to demonstrate how I integrate Testcontainers in Spring Boot tests for running integration tests with a database. I’m not using Testcontainers’ Spring Boot modules. How it works with them, I will show in a separate blog post. All samples can be found on GitHub.

Why Testcontainers?

Testcontainers is a library that helps to integrate infrastructure components like database in integration tests based on Docker Container. It helps to avoid writing integrated tests. These are kind of tests that will pass or fail based on the correctness of another system. With Testcontainers I have the control over these dependent systems.

Introducing the domain

The further samples shows different approach how to save some hero objects through different repository implementations in a database and how the corresponding tests could look like.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import javax.persistence.Entity;
import javax.persistence.GeneratedValue;
import javax.persistence.Id;
import java.util.Objects;

public class Hero {
    private Long id;
    private String name;
    private String city;
    private ComicUniversum universum;

    public Hero(String name, String city, ComicUniversum universum) {
        this.name = name;
        this.city = city;
        this.universum = universum;
    }

    public String getName() {
        return name;
    }

    public String getCity() {
        return city;
    }

    public ComicUniversum getUniversum() {
        return universum;
    }
}

All further repositories are parts of a Spring Boot web application. So at the end of this blog post I will demonstrate how to write a test for the whole web application including a database. Let’s start with an easy sample, a repository based on JDBC.

Testing Repository Based on JDBC

Assume we have following repository implementation based on JDBC. We have two methods, one for adding a hero into the database and one for getting all heroes from the database.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import org.springframework.jdbc.core.JdbcTemplate;
import org.springframework.stereotype.Repository;

import javax.sql.DataSource;
import java.util.Collection;

@Repository
public class HeroClassicJDBCRepository {

    private final JdbcTemplate jdbcTemplate;

    public HeroClassicJDBCRepository(DataSource dataSource) {
        jdbcTemplate = new JdbcTemplate(dataSource);
    }

    public void addHero(Hero hero) {
        jdbcTemplate.update("insert into hero (city, name, universum) values (?,?,?)",
                hero.getCity(), hero.getName(), hero.getUniversum().name());

    }

    public Collection<Hero> allHeros() {
        return jdbcTemplate.query("select * From hero",
                (resultSet, i) -> new Hero(resultSet.getString("name"),
                                            resultSet.getString("city"),
                                            ComicUniversum.valueOf(resultSet.getString("universum"))));
    }

}

For this repository, we can write a normal JUnit5 tests without Spring application context loading. So first at all, we have to set up the dependencies to the test libraries, in this case, JUnit5 and Testcontainers. As build tool, I use Maven. Both test libraries provide so called BOM “bill of material”, that helps to avoid a version mismatch in my used dependencies. As database, I want to use MySQL. Therefore, I use the Testcontainers’ module mysql additional to the core module testcontainers. It provides a predefined MySQL container. For simplifying the container setup specifically in JUnit5 test code, Testcontainers provides a JUnit5 module junit-jupiter.

    <dependencies>
        <dependency>
            <groupId>org.testcontainers</groupId>
            <artifactId>testcontainers</artifactId>
            <scope>test</scope>
        </dependency>
        <dependency>
            <groupId>org.testcontainers</groupId>
            <artifactId>junit-jupiter</artifactId>
            <scope>test</scope>
        </dependency>
        <dependency>
            <groupId>org.testcontainers</groupId>
            <artifactId>mysql</artifactId>
            <scope>test</scope>
        </dependency>
        <dependency>
            <groupId>org.junit.jupiter</groupId>
            <artifactId>junit-jupiter</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>
    <dependencyManagement>
        <dependencies>
            <dependency>
                <groupId>org.junit</groupId>
                <artifactId>junit-bom</artifactId>
                <version>${junit.jupiter.version}</version>
                <type>pom</type>
                <scope>import</scope>
            </dependency>
            <dependency>
                <groupId>org.testcontainers</groupId>
                <artifactId>testcontainers-bom</artifactId>
                <version>${testcontainers.version}</version>
                <type>pom</type>
                <scope>import</scope>
            </dependency>
        </dependencies>
    </dependencyManagement>

Now, we have everything to write the first test.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import ...

@Testcontainers
class HeroClassicJDBCRepositoryIT {
    @Container
    private MySQLContainer database = new MySQLContainer();

    private HeroClassicJDBCRepository repositoryUnderTest;

    @Test
    void testInteractionWithDatabase() {
        ScriptUtils.runInitScript(new JdbcDatabaseDelegate(database, ""),"ddl.sql");
        repositoryUnderTest = new HeroClassicJDBCRepository(dataSource());

        repositoryUnderTest.addHero(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));

        Collection<Hero> heroes = repositoryUnderTest.allHeros();

        assertThat(heroes).hasSize(1);
    }

    @NotNull
    private DataSource dataSource() {
        MysqlDataSource dataSource = new MysqlDataSource();
        dataSource.setUrl(database.getJdbcUrl());
        dataSource.setUser(database.getUsername());
        dataSource.setPassword(database.getPassword());
        return dataSource;
    }
}

Let’s have a look how the database is prepared for the test. Firstly, we annotate the test class with @Testcontainers. Behind this annotation hides a JUnit5 extension provided by Testcontainers. It checks if Docker is installed on the machine, starts and stops the container during the test. But how Testcontainers knows which container it should start? Here, the annotation @Container helps. It marks container that should manage by the Testcontainers extension. In this case, a MySQLContainer provided by Testcontainers module mysql. This class provides a MySQL Docker container and handles such things like setting up database user, recognizing when the database is ready to use etc. As soon as the database is ready to use, the database schema has to be set up. Testcontainers can also provide support here. ScriptUtils.runInitScript(new JdbcDatabaseDelegate(database, ""),"ddl.sql"); ensure that the schema is set up like it defines in SQL script ddl.sql.

-- ddl.sql
create table hero (id bigint AUTO_INCREMENT PRIMARY KEY, city varchar(255), name varchar(255), universum varchar(255)) engine=InnoDB

Now we are ready to set up our repository under test. Therefore, we need the database connection information for the DataSource object. Under the hood, Testcontainers searches after an available port and bind the container on this free port. This port number is different on every container start via Testcontainers. Furthermore, it configures the database in container with a user and password. Therefore, we have to ask the MySQLContainer object how the database credentials and the JDBC URL are. With this information, we can set up the repository under test (repositoryUnderTest = new HeroClassicJDBCRepository(dataSource());) and finish the test.

If you run the test and you get the following error message:

17:18:50.990 [ducttape-1] DEBUG com.github.dockerjava.core.command.AbstrDockerCmd - Cmd: org.testcontainers.dockerclient.transport.okhttp.OkHttpDockerCmdExecFactory$1@1adc57a8
17:18:51.492 [ducttape-1] DEBUG org.testcontainers.dockerclient.DockerClientProviderStrategy - Pinging docker daemon...
17:18:51.493 [ducttape-1] DEBUG com.github.dockerjava.core.command.AbstrDockerCmd - Cmd: org.testcontainers.dockerclient.transport.okhttp.OkHttpDockerCmdExecFactory$1@3e5b3a3b
17:18:51.838 [main] DEBUG org.testcontainers.dockerclient.DockerClientProviderStrategy - UnixSocketClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed). Root cause LastErrorException ([111] Verbindungsaufbau abgelehnt)
17:18:51.851 [main] DEBUG org.rnorth.tcpunixsocketproxy.ProxyPump - Listening on localhost/127.0.0.1:41039 and proxying to /var/run/docker.sock
17:18:51.996 [ducttape-0] DEBUG org.testcontainers.dockerclient.DockerClientProviderStrategy - Pinging docker daemon...
17:18:51.997 [ducttape-1] DEBUG org.testcontainers.dockerclient.DockerClientProviderStrategy - Pinging docker daemon...
17:18:51.997 [ducttape-0] DEBUG com.github.dockerjava.core.command.AbstrDockerCmd - Cmd: org.testcontainers.dockerclient.transport.okhttp.OkHttpDockerCmdExecFactory$1@5d43d23e
17:18:51.997 [ducttape-1] DEBUG com.github.dockerjava.core.command.AbstrDockerCmd - Cmd: org.testcontainers.dockerclient.transport.okhttp.OkHttpDockerCmdExecFactory$1@7abf08d2
17:18:52.002 [tcp-unix-proxy-accept-thread] DEBUG org.rnorth.tcpunixsocketproxy.ProxyPump - Accepting incoming connection from /127.0.0.1:41998
17:19:01.866 [main] DEBUG org.testcontainers.dockerclient.DockerClientProviderStrategy - ProxiedUnixSocketClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed). Root cause TimeoutException (null)
17:19:01.870 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy - Could not find a valid Docker environment. Please check configuration. Attempted configurations were:
17:19:01.872 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy -     EnvironmentAndSystemPropertyClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed)
17:19:01.873 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy -     EnvironmentAndSystemPropertyClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed)
17:19:01.874 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy -     UnixSocketClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed). Root cause LastErrorException ([111] Verbindungsaufbau abgelehnt)
17:19:01.875 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy -     ProxiedUnixSocketClientProviderStrategy: failed with exception InvalidConfigurationException (ping failed). Root cause TimeoutException (null)
17:19:01.875 [main] ERROR org.testcontainers.dockerclient.DockerClientProviderStrategy - As no valid configuration was found, execution cannot continue
17:19:01.900 [main] DEBUG 🐳 [mysql:5.7.22] - mysql:5.7.22 is not in image name cache, updating...
Mai 01, 2020 5:19:01 NACHM. org.junit.jupiter.engine.execution.JupiterEngineExecutionContext close
SEVERE: Caught exception while closing extension context: org.junit.jupiter.engine.descriptor.MethodExtensionContext@2e6a5539
org.testcontainers.containers.ContainerLaunchException: Container startup failed
	at org.testcontainers.containers.GenericContainer.doStart(GenericContainer.java:322)
	at org.testcontainers.containers.GenericContainer.start(GenericContainer.java:302)
	at org.testcontainers.junit.jupiter.TestcontainersExtension$StoreAdapter.start(TestcontainersExtension.java:173)
	at org.testcontainers.junit.jupiter.TestcontainersExtension$StoreAdapter.access$100(TestcontainersExtension.java:160)
	at org.testcontainers.junit.jupiter.TestcontainersExtension.lambda$null$3(TestcontainersExtension.java:50)
	at org.junit.jupiter.engine.execution.ExtensionValuesStore.lambda$getOrComputeIfAbsent$0(ExtensionValuesStore.java:81)
	at org.junit.jupiter.engine.execution.ExtensionValuesStore$MemoizingSupplier.get(ExtensionValuesStore.java:182)
	at org.junit.jupiter.engine.execution.ExtensionValuesStore.closeAllStoredCloseableValues(ExtensionValuesStore.java:58)
	at org.junit.jupiter.engine.descriptor.AbstractExtensionContext.close(AbstractExtensionContext.java:73)
	at org.junit.jupiter.engine.execution.JupiterEngineExecutionContext.close(JupiterEngineExecutionContext.java:53)
	at org.junit.jupiter.engine.descriptor.JupiterTestDescriptor.cleanUp(JupiterTestDescriptor.java:222)
	at org.junit.jupiter.engine.descriptor.JupiterTestDescriptor.cleanUp(JupiterTestDescriptor.java:57)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$cleanUp$9(NodeTestTask.java:151)
	at org.junit.platform.engine.support.hierarchical.ThrowableCollector.execute(ThrowableCollector.java:73)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.cleanUp(NodeTestTask.java:151)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.execute(NodeTestTask.java:83)
	at java.base/java.util.ArrayList.forEach(ArrayList.java:1540)
	at org.junit.platform.engine.support.hierarchical.SameThreadHierarchicalTestExecutorService.invokeAll(SameThreadHierarchicalTestExecutorService.java:38)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$5(NodeTestTask.java:139)
	at org.junit.platform.engine.support.hierarchical.ThrowableCollector.execute(ThrowableCollector.java:73)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$7(NodeTestTask.java:125)
	at org.junit.platform.engine.support.hierarchical.Node.around(Node.java:135)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$8(NodeTestTask.java:123)
	at org.junit.platform.engine.support.hierarchical.ThrowableCollector.execute(ThrowableCollector.java:73)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.executeRecursively(NodeTestTask.java:122)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.execute(NodeTestTask.java:80)
	at java.base/java.util.ArrayList.forEach(ArrayList.java:1540)
	at org.junit.platform.engine.support.hierarchical.SameThreadHierarchicalTestExecutorService.invokeAll(SameThreadHierarchicalTestExecutorService.java:38)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$5(NodeTestTask.java:139)
	at org.junit.platform.engine.support.hierarchical.ThrowableCollector.execute(ThrowableCollector.java:73)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$7(NodeTestTask.java:125)
	at org.junit.platform.engine.support.hierarchical.Node.around(Node.java:135)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.lambda$executeRecursively$8(NodeTestTask.java:123)
	at org.junit.platform.engine.support.hierarchical.ThrowableCollector.execute(ThrowableCollector.java:73)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.executeRecursively(NodeTestTask.java:122)
	at org.junit.platform.engine.support.hierarchical.NodeTestTask.execute(NodeTestTask.java:80)
	at org.junit.platform.engine.support.hierarchical.SameThreadHierarchicalTestExecutorService.submit(SameThreadHierarchicalTestExecutorService.java:32)
	at org.junit.platform.engine.support.hierarchical.HierarchicalTestExecutor.execute(HierarchicalTestExecutor.java:57)
	at org.junit.platform.engine.support.hierarchical.HierarchicalTestEngine.execute(HierarchicalTestEngine.java:51)
	at org.junit.platform.launcher.core.DefaultLauncher.execute(DefaultLauncher.java:229)
	at org.junit.platform.launcher.core.DefaultLauncher.lambda$execute$6(DefaultLauncher.java:197)
	at org.junit.platform.launcher.core.DefaultLauncher.withInterceptedStreams(DefaultLauncher.java:211)
	at org.junit.platform.launcher.core.DefaultLauncher.execute(DefaultLauncher.java:191)
	at org.junit.platform.launcher.core.DefaultLauncher.execute(DefaultLauncher.java:128)
	at com.intellij.junit5.JUnit5IdeaTestRunner.startRunnerWithArgs(JUnit5IdeaTestRunner.java:69)
	at com.intellij.rt.junit.IdeaTestRunner$Repeater.startRunnerWithArgs(IdeaTestRunner.java:33)
	at com.intellij.rt.junit.JUnitStarter.prepareStreamsAndStart(JUnitStarter.java:230)
	at com.intellij.rt.junit.JUnitStarter.main(JUnitStarter.java:58)
Caused by: org.testcontainers.containers.ContainerFetchException: Can't get Docker image: RemoteDockerImage(imageNameFuture=java.util.concurrent.CompletableFuture@539d019[Completed normally], imagePullPolicy=DefaultPullPolicy(), dockerClient=LazyDockerClient.INSTANCE)
	at org.testcontainers.containers.GenericContainer.getDockerImageName(GenericContainer.java:1265)
	at org.testcontainers.containers.GenericContainer.logger(GenericContainer.java:600)
	at org.testcontainers.containers.GenericContainer.doStart(GenericContainer.java:311)
	... 47 more
Caused by: java.lang.IllegalStateException: Previous attempts to find a Docker environment failed. Will not retry. Please see logs and check configuration
	at org.testcontainers.dockerclient.DockerClientProviderStrategy.getFirstValidStrategy(DockerClientProviderStrategy.java:78)
	at org.testcontainers.DockerClientFactory.client(DockerClientFactory.java:115)
	at org.testcontainers.LazyDockerClient.getDockerClient(LazyDockerClient.java:14)
	at org.testcontainers.LazyDockerClient.inspectImageCmd(LazyDockerClient.java:12)
	at org.testcontainers.images.LocalImagesCache.refreshCache(LocalImagesCache.java:42)
	at org.testcontainers.images.AbstractImagePullPolicy.shouldPull(AbstractImagePullPolicy.java:24)
	at org.testcontainers.images.RemoteDockerImage.resolve(RemoteDockerImage.java:62)
	at org.testcontainers.images.RemoteDockerImage.resolve(RemoteDockerImage.java:25)
	at org.testcontainers.utility.LazyFuture.getResolvedValue(LazyFuture.java:20)
	at org.testcontainers.utility.LazyFuture.get(LazyFuture.java:27)
	at org.testcontainers.containers.GenericContainer.getDockerImageName(GenericContainer.java:1263)
	... 49 more



org.testcontainers.containers.ContainerLaunchException: Container startup failed

This error message means that the Docker daemon is not running. After ensuring that Docker daemon is running, the test run is successful.

There are very many debug messages in the console output. The logging output in tests can be configured by a logback.xml file in src/test/resources:

<?xml version="1.0" encoding="UTF-8" ?>
<configuration>
    <include resource="org/springframework/boot/logging/logback/base.xml"/>
    <root level="info">
        <appender-ref ref="CONSOLE" />
    </root>
</configuration>

Spring Boot documentation about logging recommends to use logback-spring.xml as configuration file. But normal JUnit5 tests don’t recognize it, only @SpringBootTest annotated tests. logback.xml is used by both kind of tests.

Testing Repository based on JPA Entity Manager

Now, we want to implement a repository based on JPA with a classic entity manager. Assume, we have following implementation with three methods, adding heroes to the database, finding heroes by search criteria and getting all heroes from the database. The entity manager is configured by Spring’s application context (@PersistenceContext is responsible for that).

package com.github.sparsick.testcontainerspringboot.hero.universum;

import ...

@Repository
public class HeroClassicJpaRepository {

    @PersistenceContext
    private EntityManager em;

    @Transactional
    public void addHero(Hero hero) {
        em.persist(hero);
    }

    public Collection<Hero> allHeros() {
        return em.createQuery("Select hero FROM Hero hero", Hero.class).getResultList();
    }

    public Collection<Hero> findHerosBySearchCriteria(String searchCriteria) {
        return em.createQuery("SELECT hero FROM Hero hero " +
                        "where hero.city LIKE :searchCriteria OR " +
                        "hero.name LIKE :searchCriteria OR " +
                        "hero.universum = :searchCriteria",
                Hero.class)
                .setParameter("searchCriteria", searchCriteria).getResultList();
    }

}

As JPA implementation, we choose Hibernate and MySQL as database provider. We have to configure which dialect should Hibernate use.

# src/main/resources/application.properties
spring.jpa.database-platform=org.hibernate.dialect.MySQL8Dialect

In application.properties you also configure the database connection etc.

For setting up the entity manager in a test correctly, we have to run the test with an application context, so that entity manager is configured correctly by Spring.

Spring Boot brings some test support classes. Therefore, we have to add a further test dependency to the project.

        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-test</artifactId>
            <scope>test</scope>
        </dependency>

This starter also includes JUnit Jupiter dependency and dependencies from other test library, so you can remove these dependencies from your dependency declaration if you want to.

Now, we have everything for writing the test.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import ...

@SpringBootTest
@Testcontainers
class HeroClassicJpaRepositoryIT {
    @Container
    private static MySQLContainer database = new MySQLContainer();

    @Autowired
    private HeroClassicJpaRepository repositoryUnderTest;

    @Test
    void findHeroByCriteria(){
        repositoryUnderTest.addHero(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));

        Collection<Hero> heros = repositoryUnderTest.findHerosBySearchCriteria("Batman");

        assertThat(heros).contains(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));
    }
   
    @DynamicPropertySource
    static void databaseProperties(DynamicPropertyRegistry registry) {
        registry.add("spring.datasource.url", database::getJdbcUrl);
        registry.add("spring.datasource.username", database::getUsername);
        registry.add("spring.datasource.password", database::getPassword);
    }
}

The test class is annotated with some annotations. The first one is @SpringBootTest thereby the Spring application context is started during the test. The last one is @Testcontainers . This annotation we already know from the last test. It is a JUnit5 extension that manage starting and stopping the docker container during the test. Like we see at above JDBC test, we annotate database container private static MySQLContainer database = new MySQLContainer(); with @Container. It marks that this container should be managed by Testcontainers. Here is a little difference to above JDBC set up. Here, MySQLContainer database is static and in the JDBC set up it is a normal class field. Here, it has to be static because the container has to start before the application context starts, so that we have a change to pass the database connection configuration to the application context. For this, static method databaseProperties is responsible. Here, it is important that this method is annotated by @DynamicPropertySource. It overrides the application context configuration during the start phase. In our case, we want to override the database connection configuration with the database information that we get from database container object managed by Testcontainers. The last step is to set up the database schema in the database. Here JPA can help. It can create a database schema automatically. You have to configure it with

# src/test/resources/application.properties
spring.jpa.hibernate.ddl-auto=update

Now, we can inject the repository into the test (@Autowired private HeroClassicJpaRepository repositoryUnderTest). This repository is configured by Spring and ready to test.

Testing Repository based on Spring Data JPA

Today, it is common in a Spring Boot application to use JPA in combination with Spring Data, so we rewrite our repository to use Spring Data JPA instead of plain JPA. The result is an interface that extends Spring Data’s CrudRepository, so we have all basic operation like save, delete, update find by id etc. . For searching by criteria functionality, we have to define a method with @Query annotation that have a JPA query.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import org.springframework.data.jpa.repository.Query;
import org.springframework.data.repository.CrudRepository;
import org.springframework.data.repository.query.Param;

import java.util.List;

public interface HeroSpringDataJpaRepository extends CrudRepository<Hero, Long> {

    @Query("SELECT hero FROM Hero hero where hero.city LIKE :searchCriteria OR hero.name LIKE :searchCriteria OR hero.universum = :searchCriteria")
    List<Hero> findHerosBySearchCriteria(@Param("searchCriteria") String searchCriteria);
}

As mentioned above in classic JPA sample so also here, we have to configure which SQL dialect our chosen JPA implementation Hibernate should use and how the database schema should set up.

The same with the test configuration, again we need a test with a Spring application context to configure the repository correctly for the test. But here we don’t need to start the whole application context with @SpringBootTest. Instead, we use @DataJpaTest. This annotation starts an application context only with beans that are needed for the persistence layer.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import ...

@DataJpaTest
@AutoConfigureTestDatabase(replace = AutoConfigureTestDatabase.Replace.NONE)
@Testcontainers
class HeroSpringDataJpaRepositoryIT {
    @Container
    private static MySQLContainer database = new MySQLContainer();

    @Autowired
    private HeroSpringDataJpaRepository repositoryUnderTest;

    @Test
    void findHerosBySearchCriteria() {
        repositoryUnderTest.save(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));

        Collection<Hero> heros = repositoryUnderTest.findHerosBySearchCriteria("Batman");

        assertThat(heros).hasSize(1).contains(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));
    }

    @DynamicPropertySource
    static void databaseProperties(DynamicPropertyRegistry registry) {
        registry.add("spring.datasource.url",database::getJdbcUrl);
        registry.add("spring.datasource.username", database::getUsername);
        registry.add("spring.datasource.password", database::getPassword);
    }
}

@DataJpaTest starts an in-memory database as default. But we want that a containerized database is used, provided by Testcontainers. Therefore, we have to add the annotation @AutoConfigureTestDatabase(replace = AutoConfigureTestDatabase.Replace.NONE). This disables starting an in-memory database. The remaining test configuration is the same as the configuration in above test for the plain JPA example.

Testing Repositories but Reusing a database

With the increasing number of tests, it becomes more and more important that each test takes quite a long time, because each time a new database is started and initialized. One idea is to reuse the database in each test. Here the Single Container Pattern can help. A database is started and initialized once before all tests start running. For that, each test that need a database has to extend an abstract class, that is responsible for starting and initializing a database once before all tests run.

package com.github.sparsick.testcontainerspringboot.hero.universum;

import ...


public abstract class DatabaseBaseTest {
    static final MySQLContainer DATABASE = new MySQLContainer();

    static {
        DATABASE.start();
    }

    @DynamicPropertySource
    static void databaseProperties(DynamicPropertyRegistry registry) {
        registry.add("spring.datasource.url", DATABASE::getJdbcUrl);
        registry.add("spring.datasource.username", DATABASE::getUsername);
        registry.add("spring.datasource.password", DATABASE::getPassword);
    }
}

In this abstract class we configure the database that is started once for all tests that extend this abstract class and the application context with that database. Please note, that we don’t use Testcontainers’ annotations here, because this annotation takes care that the container is started and stopped after each test. But this we would avoid. Therefore, we start the database by ourselves. For stopping database we don’t need to take care. For this Testcontainers’ side-car container ryuk takes care.

Now, each test class, that need a database, extends this abstract class. The only thing, that we have to configure, is how the application context should be initialized. That means, when you need the whole application context then use @SpringBootTest. When you need only persistence layer then use @DataJpaTest with @AutoConfigureTestDatabase(replace = AutoConfigureTestDatabase.Replace.NONE).

@DataJpaTest
@AutoConfigureTestDatabase(replace = AutoConfigureTestDatabase.Replace.NONE)
class HeroSpringDataJpaRepositoryReuseDatabaseIT extends DatabaseBaseTest {

    @Autowired
    private HeroSpringDataJpaRepository repositoryUnderTest;

    @Test
    void findHerosBySearchCriteria() {
        repositoryUnderTest.save(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));

        Collection<Hero> heros = repositoryUnderTest.findHerosBySearchCriteria("Batman");

        assertThat(heros).contains(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));
    }
}

Testing the whole Web Application including Database

Now we want to test our whole application, from controller to database. The controller implementation looks like this:

@RestController
public class HeroRestController {

    private final HeroSpringDataJpaRepository heroRepository;

    public HeroRestController(HeroSpringDataJpaRepository heroRepository) {
        this.heroRepository = heroRepository;
    }

    @GetMapping("heros")
    public Iterable<Hero> allHeros(String searchCriteria) {
        if (searchCriteria == null || searchCriteria.equals("")) {
            return heroRepository.findAll();

        }
        return heroRepository.findHerosBySearchCriteria(searchCriteria);
    }

    @PostMapping("hero")
    public void hero(@RequestBody Hero hero) {
        heroRepository.save(hero);
    }
}

The test class that test the whole way from database to controller looks like that

@SpringBootTest
@AutoConfigureMockMvc
@Testcontainers
class HeroRestControllerIT {

    @Container
    private static MySQLContainer database = new MySQLContainer();

    @Autowired
    private MockMvc mockMvc;

    @Autowired
    private HeroSpringDataJpaRepository heroRepository;

    @Test
    void allHeros() throws Exception {
        heroRepository.save(new Hero("Batman", "Gotham City", ComicUniversum.DC_COMICS));
        heroRepository.save(new Hero("Superman", "Metropolis", ComicUniversum.DC_COMICS));

        mockMvc.perform(get("/heros"))
                .andExpect(status().isOk())
                .andExpect(jsonPath("$[*].name", containsInAnyOrder("Batman", "Superman")));
    }

    @DynamicPropertySource
    static void databaseProperties(DynamicPropertyRegistry registry) {
        registry.add("spring.datasource.url", database::getJdbcUrl);
        registry.add("spring.datasource.username", database::getUsername);
        registry.add("spring.datasource.password", database::getPassword);
    }
}

The test set up for the database and the application is known by the test from the above sections. One thing is different. We add MockMVC support with @AutoConfigureMockMvc. This helps to write tests through the HTTP layer.

Of course, you can also use the single container pattern in which the abstract class DatabaseBaseTest is extended.

Conclusion and Overview

This blog post shows how we can write tests for some persistence layer implementations in Spring Boot with Testcontainers. We also see how to reuse database instance for several tests and how to write test for the whole web application from controller tor database. All code snippet can be found on GitHub. In a further blog post I will show how to write test with Testcontainers Spring Boot modules.

Do you have other ideas for writing tests for persistence layer? Please let me know and write a comment.

Further Information

  1. Concept of BOM “bill of material”
  2. Testcontainers
  3. Spring Boot Documentation – Logging
  4. Spring Boot Documentation – Auto-configured Data JPA Tests
  5. Testcontainers – Single Container Pattern
  6. Spring Boot Documentation – MockMVC
  7. Full example in GitHub repository


Leave a comment

Test Environment for Ansible on a Windows System Without Linux Subsystem Support

Some weeks ago, I gave a workshop about Ansible and I was asked how to set up a local test environment for Ansible. Requirement is that this test environment can run on a Windows 10 without a Linux subsystem. I don’t why, but it was not my first customer where Windows 10 could not be used with Linux subsystem (Don’t ask me about reasons). So I recommend following set up.

Idea

Ansible itself requires a Linux-based system as the control machine (so-called control node). The machine that is provisioned (so-called managed node) can be both, Linux- or Windows-based. In my following example, the manage node is Linux-based.   So when I have to develop on a Windows machine, I install two Linux-based virtual machines. One for calling the Ansible’s playbooks and a second one for testing the provisioning. I set up the virtual machines with VirtualBox and Vagrant. Vagrant allows me to share the playbooks easily between host and virtual machines. That means, I can develop the playbook on the Windows system and the virtual machines can run headless. The next section shows you how to set up this tool chain.

Tool Chain Setup

At first, install VirtualBox and Vagrant on your machine. I additionally use Babun, a windows shell based on Cygwin and oh-my-zsh, for a better shell experience on Windows, but this isn’t necessary. Then, go to the directory (let’s called it ansible-workspace), where your Ansible’s playbooks are located. Create there a Vagrant configuration file with the command vagrant init:

$ cd ansible-workspace
$ vagrant init
$ tree  
.
├── deploy-app.yml
├── host_vars
│   └── ubuntu-server
├── inventories
│   ├── production
│   └── test
├── roles
│   ├── deploy-on-tomcat
│   │   ├── defaults
│   │   │   └── main.yml
│   │   └── tasks
│   │       ├── cleanup-webapp.yml
│   │       ├── deploy-webapp.yml
│   │       ├── main.yml
│   │       ├── start-tomcat.yml
│   │       └── stop-tomcat.yml
└── Vagrantfile

Now, we have to choose a so-called Vagrant Box on Vagrant Cloud. A box is the package format for a Vagrant environment. It depends on the provider and the operating system that you choose to use. In our case, it is a VirtualBox VM image based on a minimal Ubuntu 18.04 system (box name is bento/ubuntu-18.04 ). In our case, we have to configure two boxes in our Vagrantfile. Both are based on bento/ubuntu-18.04.

Vagrant.configure("2") do |config|  
  config.vm.define "managed-node" do |ubuntu|
    ubuntu.vm.box = "bento/ubuntu-18.04"
  end

  config.vm.define "control-node" do |vbox|
     vbox.vm.box = "bento/ubuntu-18.04"
  end
end

The next step is to ensure that Python is installed on the managed node and this node is available via an IP address. For that, we use the shell provisioning, and we configure a private network in the Vagrantfile:

# ...  
  config.vm.define "managed-node" do |ubuntu|
    ubuntu.vm.box = "bento/ubuntu-18.04"
    ubuntu.vm.network "private_network", ip: "192.168.33.11"

    ubuntu.vm.provision "shell", inline: <<-SHELL
      sudo apt-get update
      sudo apt-get install python -y
    SHELL
  end
# ...

In the control node, we have to ensure that Ansible is installed and that a SSH connection from control node to our managed node via SSH-key is possible. For that agian, we use the shell provisioning in the Vagrantfile:

 # ... 
  config.vm.define "controll-node" do |vbox|
     vbox.vm.box = "bento/ubuntu-18.04"

     vbox.vm.provision "shell", inline: <<-SHELL
       sudo apt-get update -y
       sudo apt-get install -y software-properties-common
       sudo apt-add-repository ppa:ansible/ansible
       sudo apt-get update -y
       sudo apt-get install -y ansible
     SHELL

     vbox.vm.provision "shell", privileged: false, inline: <<-SHELL
       ssh-keygen -t rsa -q -f "/home/vagrant/.ssh/id_rsa" -N ""
       ssh-keygen -R 192.168.33.11
       ssh-keyscan -t rsa -H 192.168.33.11 >> /home/vagrant/.ssh/known_hosts
       sshpass -p 'vagrant' ssh-copy-id -i /home/vagrant/.ssh/id_rsa.pub vagrant@192.168.33.11
     SHELL
  end
# ...

On Github’s Gist you can find the whole Vagrantfile.

The last step is to configure your Ansible project, so that the playbooks can run against the managed node. Therefore, you create a new inventory file called local-test with managed node’s IP address:

$ cat inventories/local-test
[application_server] # name of your group that is mention in your playbook
192.168.33.11

Workflow

After setting up the tool chain let’s have a look how to work with it. I write my Ansible playbook on the Windows system and run them from the control node against managed node. First at all, we have to start our Vagrant boxes.

$ cd ansible-workspace
$ vagrant up

When both Vagrant boxes are ready to use, we can jump into control node box with:

$ vagrant ssh control-node

You can find the Ansible playbooks inside the box in the folder /vagrant .  In this folder run Ansible:

$ cd /vagrant
$ ansible-playbook -i inventories/local-test -u vagrant deploy-app.yml

It is important that you use the local-test inventory.

Outlook

Some Docker fans would prefer a container instead of a virtual machine. But remember, Docker runs on Windows in a virtual machine, therefore, I don’t see a benefit for using Docker instead of a virtual machine. But of course with Windows 10 native container support a setup with Docker is a good alternative if Ansible doesn’t run on the Linux subsystem.

Do you another idea or approach? Let me know and write a comment.

Links

  1. VirtualBox
  2. Vagrant
  3. Whole Vagrantfile on GitHub.


Leave a comment

Software Engineering Camp 2019

Some weeks ago, I visited the bar camp “Software Engineering Camp 2019” in Erlangen. It was my first time visiting. I’d like to share my experience.

Friday evening

The bar camp started on Friday evening with three lighting talks. Firstly, Markus Harrer talked about psychology in software development.

The second lightning talk by Sasabe-san was about the mindset behind ShuHaRi. He is a Japanese software engineer, and he explained what the Japanese understand by ShuHaRi. And surprise, surprise, they have a different understanding than Europeans.

The last lightning talk by Markus, founder of OwnCloud, told the story how SUSE creates Enterprise Linux.

Saturday

The Open Space Marketplace started after a great breakfast. Here my notes from some session I visited on Saturday.

Modern Linux CLI Tools

Frank and Patrick did a great session about Linux CLI tools. I learned some new cool tools.

My notes

Patrick prepared some “slides”. The presentation runs within a shell, very geeky.

Sketch notes from other participants:

CSS Needs More Software Craftsmanship

My second session was about why CSS code needs more attention. We talked about how we can improve the readability and maintainability of CSS code.

Boring Software Manifesto

I hosted together with Dominik a session about building software with a “boring” technology and why it could be a good idea.

Whiteboard notes from the session

Awesome sketch note from Nora:

From Developer to Coporated Learning Coach

Latti told his story how he got a learning coach. He told us how the software craftsmanship community changes his life.

My notes

Infrastructure As Code With Terraform

Nora gave a good introduction to Terraform.

My notes

Saturday Evening Activities

At the evening, we had some very geeky activities, like Hacker Jeopardy, Retro Gaming, Craftbeer tasting.

I participated in the first round of Hacker-Jeopardy. I lose 🙂

My last social activity was some rounds of Singstar :).

Sunday

Unfortunately, I had to leave the bar camp after breakfast. So I missed the morning session. But I could follow on my timeline on Twitter, that I missed some very good sessions like

ShuHaRi

Software Engineering and Ethics

Sketch Note Session

Conclusion

For me, it was an amazing bar camp. I can recommend it.


Leave a comment

Strategy Pattern Revisited With Spring

This blog post wants to show another approach how to implement the Strategy Pattern with dependency injection. As DI framework, I choose Spring framework

From Wikipedia

Firstly, let’s have a look how the Strategy Pattern is implemented in the classic way.
As starting point, we have a HeroController that should add a hero in HeroRepository depends on which repository was chosen by the user.

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Controller;
import org.springframework.web.bind.annotation.ModelAttribute;
import org.springframework.web.bind.annotation.PostMapping;

@Controller
public class HeroControllerClassicWay {

    @PostMapping("/hero/new")
    public String addNewHero(@ModelAttribute("newHero") NewHeroModel newHeroModel) {
        HeroRepository heroRepository = findHeroRepository(newHeroModel.getRepository());
        heroRepository.addHero(newHeroModel.getHero());
        return "redirect:/hero";
    }

    private HeroRepository findHeroRepository(String repositoryName) {
        if (repositoryName.equals("Unique")) {
            return new UniqueHeroRepository();
        }

        if(repositoryName.equals(("Duplicate")){
            return new DuplicateHeroRepository();
        }

        throw new IllegalArgumentException(String.format("Find no repository for given repository name [%s]", repositoryName));
    }
}
package com.github.sparsick.springbootexample.hero.universum;

import java.util.Collection;
import java.util.HashSet;
import java.util.Set;

import org.springframework.stereotype.Repository;

@Repository
public class UniqueHeroRepository implements HeroRepository {

    private Set<Hero> heroes = new HashSet<>();

    @Override
    public String getName() {
        return "Unique";
    }

    @Override
    public void addHero(Hero hero) {
        heroes.add(hero);
    }

    @Override
    public Collection<Hero> allHeros() {
        return new HashSet<>(heroes);
    }

}
package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Repository;

import java.util.ArrayList;
import java.util.Collection;
import java.util.List;

@Repository
public class DuplicateHeroRepository implements HeroRepository {

    private List<Hero> heroes = new ArrayList<>();

    @Override
    public void addHero(Hero hero) {
        heroes.add(hero);
    }

    @Override
    public Collection<Hero> allHeros() {
        return List.copyOf(heroes);
    }

    @Override
    public String getName() {
        return "Duplicate";
    }
}

This implementation has some pitfalls. The creation of the repository implementations aren’t managed by the Spring Context (it breaks the dependency injection / inverse of control). This will be painful as soon as you want to expand the repository implementation with further feature that need to inject other classes (for example, counting the usage of this class with MeterRegistry).

package com.github.sparsick.springbootexample.hero.universum;

import java.util.Collection;
import java.util.HashSet;
import java.util.Set;

import io.micrometer.core.instrument.Counter;
import io.micrometer.core.instrument.MeterRegistry;
import org.springframework.stereotype.Repository;

@Repository
public class UniqueHeroRepository implements HeroRepository {

    private Set<Hero> heroes = new HashSet<>();
    private Counter addCounter;

    public UniqueHeroRepository(MeterRegistry meterRegistry) {
        addCounter = meterRegistry.counter("hero.repository.unique");
    }

    @Override
    public String getName() {
        return "Unique";
    }

    @Override
    public void addHero(Hero hero) {
        addCounter.increment();
        heroes.add(hero);
    }

    @Override
    public Collection<Hero> allHeros() {
        return new HashSet<>(heroes);
    }

}

It breaks also the separation of concern. When I want to test the controller class, I have no possibility to mock the repository interface easily. So the first idea is to put the creation of repository implementation to the Spring context. The repository implementation are annotated with @Repository annotation. So Spring’s component scan find them.
The next question how to inject them into the controller class. Here, a Spring feature can help. I define a list of HeroRepository in the controller. This list has to be filled during the creation of the controller instance.

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Controller;
import org.springframework.web.bind.annotation.ModelAttribute;
import org.springframework.web.bind.annotation.PostMapping;

import java.util.List;

@Controller
public class HeroControllerRefactoringStep1 {

    private List<HeroRepository> heroRepositories;

    public HeroControllerRefactoringStep1(List<HeroRepository> heroRepositories) {
        this.heroRepositories = heroRepositories;
    }

    @PostMapping("/hero/new")
    public String addNewHero(@ModelAttribute("newHero") NewHeroModel newHeroModel) {
        HeroRepository heroRepository = findHeroRepository(newHeroModel.getRepository());
        heroRepository.addHero(newHeroModel.getHero());
        return "redirect:/hero";
    }

    private HeroRepository findHeroRepository(String repositoryName) {
        return heroRepositories.stream()
                .filter(heroRepository -> heroRepository.getName().equals(repositoryName))
                .findFirst()
                .orElseThrow(()-> new IllegalArgumentException(String.format("Find no repository for given repository name [%s]", repositoryName)));

    }
}

Spring searches in its context for all implementation of the interface HeroRepostiory and put them all to the list. One disadvantage has this solution, every adding a hero browses the list of HeroRepository to find the right implementation. This can be optimized by creating a map in the controller constructor that has the repository name as key and the corresponded implementation as value.

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Controller;
import org.springframework.web.bind.annotation.ModelAttribute;
import org.springframework.web.bind.annotation.PostMapping;

import java.util.HashMap;
import java.util.List;
import java.util.Map;

@Controller
public class HeroControllerRefactoringStep2 {

    private Map<String, HeroRepository> heroRepositories;

    public HeroControllerRefactoringStep2(List<HeroRepository> heroRepositories) {
        this.heroRepositories = heroRepositoryStrategies(heroRepositories);
    }

    private Map<String, HeroRepository> heroRepositoryStrategies(List<HeroRepository> heroRepositories){
        Map<String, HeroRepository> heroRepositoryStrategies = new HashMap<>();
        heroRepositories.forEach(heroRepository -> heroRepositoryStrategies.put(heroRepository.getName(), heroRepository));
        return heroRepositoryStrategies;
    }

    @PostMapping("/hero/new")
    public String addNewHero(@ModelAttribute("newHero") NewHeroModel newHeroModel) {
        HeroRepository heroRepository = findHeroRepository(newHeroModel.getRepository());
        heroRepository.addHero(newHeroModel.getHero());
        return "redirect:/hero";
    }

    private HeroRepository findHeroRepository(String repositoryName) {
        HeroRepository heroRepository = heroRepositories.get(repositoryName);
        if(heroRepository != null) {
            return  heroRepository;
        }
        throw new IllegalArgumentException(String.format("Find no repository for given repository name [%s]", repositoryName));
    }
}

The final question is what if other classes in the application need the possibility to choose a repository implementation during the runtime. I could copy and paste the private method in each class that have this need or I move the creation of the map to the Spring Context and inject the Map to each class.

package com.github.sparsick.springbootexample.hero;

import com.github.sparsick.springbootexample.hero.universum.HeroRepository;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;
import org.springframework.context.annotation.Bean;

import java.util.HashMap;
import java.util.List;
import java.util.Map;

@SpringBootApplication
public class HeroApplicationRefactoringStep3 {

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

	@Bean
    Map<String, HeroRepository> heroRepositoryStrategy(List<HeroRepository> heroRepositories){
        Map<String, HeroRepository> heroRepositoryStrategy = new HashMap<>();
        heroRepositories.forEach(heroRepository -> heroRepositoryStrategy.put(heroRepository.getName(), heroRepository));
        return heroRepositoryStrategy;
    }
}

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Controller;
import org.springframework.ui.Model;
import org.springframework.web.bind.annotation.ModelAttribute;
import org.springframework.web.bind.annotation.PostMapping;

import java.util.Map;

@Controller
public class HeroControllerRefactoringStep3 {

    private Map<String, HeroRepository> heroRepositoryStrategy;

    public HeroControllerRefactoringStep3(Map<String, HeroRepository> heroRepositoryStrategy) {
        this.heroRepositoryStrategy = heroRepositoryStrategy;
    }

    @PostMapping("/hero/new")
    public String addNewHero(@ModelAttribute("newHero") NewHeroModel newHeroModel) {
        HeroRepository heroRepository = findHeroRepository(newHeroModel.getRepository());
        heroRepository.addHero(newHeroModel.getHero());
        return "redirect:/hero";
    }

    private HeroRepository findHeroRepository(String repositoryName) {
        return heroRepositoryStrategy.get(repositoryName);
    }

}

This solution is a little bit ugly, because it isn’t obvious that the Strategy Pattern is used. So the next refactoring step is moving the map of hero repositories to an own component class. Therefore, the bean definition heroRepositoryStrategy in the application configuration can be removed.

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Component;

import java.util.Collection;
import java.util.HashMap;
import java.util.Map;
import java.util.Set;

@Component
public class HeroRepositoryStrategy {

    private Map<String, HeroRepository> heroRepositoryStrategies;

    public HeroRepositoryStrategy(Set<HeroRepository> heroRepositories) {
        heroRepositoryStrategies = createStrategies(heroRepositories);
    }

    HeroRepository findHeroRepository(String repositoryName) {
        return heroRepositoryStrategies.get(repositoryName);
    }

    Set<String> findAllHeroRepositoryStrategyNames () {
        return heroRepositoryStrategies.keySet();
    }

    Collection<HeroRepository> findAllHeroRepositories(){
        return heroRepositoryStrategies.values();
    }


    private Map<String, HeroRepository> createStrategies(Set<HeroRepository> heroRepositories){
        Map<String, HeroRepository> heroRepositoryStrategies = new HashMap<>();
        heroRepositories.forEach(heroRepository -> heroRepositoryStrategies.put(heroRepository.getName(), heroRepository));
        return heroRepositoryStrategies;
    }

}

package com.github.sparsick.springbootexample.hero.universum;

import org.springframework.stereotype.Controller;
import org.springframework.ui.Model;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.ModelAttribute;
import org.springframework.web.bind.annotation.PostMapping;

import java.net.Inet4Address;
import java.net.UnknownHostException;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;

@Controller
public class HeroController {

    private HeroRepositoryStrategy heroRepositoryStrategy;

    public HeroController(HeroRepositoryStrategy heroRepositoryStrategy) {
        this.heroRepositoryStrategy = heroRepositoryStrategy;
    }

    @PostMapping("/hero/new")
    public String addNewHero(@ModelAttribute("newHero") NewHeroModel newHeroModel) {
        HeroRepository heroRepository = heroRepositoryStrategy.findHeroRepository(newHeroModel.getRepository());
        heroRepository.addHero(newHeroModel.getHero());
        return "redirect:/hero";
    }
}

The whole sample is hosted on GitHub.


Leave a comment

Maven Project Setup for Mixing Spock 1.x and JUnit 5 Tests

I create a sample Groovy project for Maven, that mixes Spock tests and JUnit 5 tests in one project. In the next section I’ll describe how to set up such kind of Maven project.

Enable Groovy in the Project

First at all, you have to enable Groovy in your project. One possibility is to add the GMavenPlus Plugin to your project.

<build>
    <plugins>
        <plugin>
            <groupId>org.codehaus.gmavenplus</groupId>
            <artifactId>gmavenplus-plugin</artifactId>
            <version>1.6.2</version>
            <executions>
                <execution>
                    <goals>
                        <goal>addSources</goal>
                        <goal>addTestSources</goal>
                        <goal>compile</goal>
                        <goal>compileTests</goal>
                    </goals>
                </execution>
            </executions>
        </plugin>
    </plugins>
</build>

The goals addSources and addTestSources add Groovy (test) sources to Maven’s main (test) sources. The default locations are src/main/groovy (for main source) and src/test/groovy (for test source). Goals compile and compileTests compile the Groovy (test) code. If you don’t have Groovy main code, you can omit addSource and compile.

This above configuration is always using the latest released Groovy version. If you want to ensure that a specific Groovy version is used, you have to add the specific Groovy dependency to your classpath.

   <dependencies>
        <dependency>
            <groupId>org.codehaus.groovy</groupId>
            <artifactId>groovy</artifactId>
            <version>2.5.6</version>
        </dependency>
  </dependencies>

Enable JUnit 5 in the Project

The simplest setup for using JUnit 5 in your project is to add the JUnit Jupiter dependency in your test class path and to configure the correct version of Maven Surefire Plugin (at least version 2.22.0).

    <dependencies>
<!--... maybe more dependencies -->
        <dependency>
            <groupId>org.junit.jupiter</groupId>
            <artifactId>junit-jupiter</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

    <dependencyManagement>
        <dependencies>
            <dependency>
                <groupId>org.junit</groupId>
                <artifactId>junit-bom</artifactId>
                <version>${junit.jupiter.version}</version>
                <scope>import</scope>
                <type>pom</type>
            </dependency>
        </dependencies>
    </dependencyManagement>
    <build>
        <plugins>
        <!-- other plugins -->
            <plugin>
                <artifactId>maven-surefire-plugin</artifactId>
                <version>2.22.1</version>
            </plugin>
        </plugins>
    </build>

Enable Spock in the Project

Choosing the right Spock dependency depends on which Groovy version you are using in the project. In our case, a Groovy version 2.5. So we need Spock in version 1.x-groovy-2.5 in our test class path.

    <dependencies>
        <!-- more dependencies -->
        <dependency>
            <groupId>org.spockframework</groupId>
            <artifactId>spock-core</artifactId>
            <version>1.3-groovy-2.5</version>
            <scope>test</scope>
        </dependency>
    </dependencies>

Now the expectation is that the Spock tests and the JUnit5 tests are executed in the Maven build. But only the JUnit5 tests are executed by Maven. So what happened?

I started to change the Maven Surefire Plugin version to 2.21.0. Then the Spock tests were executed, but no JUnit5 tests. The reason is that in the version 2.22.0 of Maven Surefire Plugin JUnit4 provider is replaced by JUnit Platform Provider as default. But Spock in version 1.x is based on JUnit4. This will be changed in Spock version 2. This version will be based on the JUnit5 Platform. Thus, for Spock 1.x, we have to add JUnit Vintage dependency to our test class path.

    <dependencies>
        <!-- more dependencies -->
          <dependency>  <!--Only necessary for surefire to run spock tests during the maven build -->
            <groupId>org.junit.vintage</groupId>
            <artifactId>junit-vintage-engine</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

This allows running elder JUnit (3/4) tests on the JUnit Platform. With this configuration both, Spock and JUnit 5 tests, are executed in the Maven build.

Links


Leave a comment

Using JUnit 5 In Pre-Java 8 Projects

This post demonstrates how JUnit 5 can be used in pre-Java 8 projects and explains why it could be a good idea.

JUnit 5 requires at least Java 8 as runtime environment, so you want to update your whole project to Java 8. But sometimes there exists reason why you can’t immediately update your project to Java 8. For example, the version of your application server in production only supports Java 7. But an update isn’t be taken quickly because of some issues in your production code.

Now, the question is how can you use JUnit 5 without update your production code to Java 8?

You can set up the Java version separately for production code and for test code .

<!-- in Maven -->
<build>
<plugins>
<plugin>
<artifactId>maven-compiler-plugin</artifactId>
<configuration>
<source>7</source>
<target>7</target>
<testSource>8</testSource>
<testTarget>8</testTarget>
</configuration>
</plugin>
</plugins>
</build>
// in Gradle
sourceCompatibility = '7'
targetCompatibility = '7'

compileTestJava {
sourceCompatibility ='8'
targetCompatibility = '8'
}

Precondition is that you use a Java 8 JDK for your build.

If you try to use Java 8 feature in your Java 7 production code, Maven and Gradle will fail the build.

// Maven
[ERROR] Failed to execute goal org.apache.maven.plugins:maven-compiler-plugin:3.8.0:compile (default-compile) on project junit5-in-pre-java8-projects: Compilation failure
[ERROR] /home/sparsick/dev/workspace/junit5-example/junit5-in-pre-java8-projects/src/main/java/Java7Class.java:[8,58] lambda expressions are not supported in -source 7
[ERROR]   (use -source 8 or higher to enable lambda expressions)
// Gradle
> Task :compileJava FAILED
/home/sparsick/dev/workspace/junit5-example/junit5-in-pre-java8-projects/src/main/java/Java7Class.java:8: error: lambda expressions are not supported in -source 7
Function<String, String > java8Feature = (input) -> input;
^
(use -source 8 or higher to enable lambda expressions)
1 error

FAILURE: Build failed with an exception.

* What went wrong:
Execution failed for task ':compileJava'.
> Compilation failed; see the compiler error output for details.

Now you can introduce JUnit 5 in your project and start writing test with JUnit 5.

<!-- Maven-->
<dependency>
<groupId>org.junit.jupiter</groupId>
<artifactId>junit-jupiter-api</artifactId>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.junit.jupiter</groupId>
<artifactId>junit-jupiter-engine</artifactId>
<scope>test</scope>
</dependency>
<dependency>
<groupId>org.junit.jupiter</groupId>
<artifactId>junit-jupiter-params</artifactId>
<scope>test</scope>
</dependency>
<!-- junit-vintage-engine is needed for running elder JUnit4 test with JUnit5-->
<dependency>
<groupId>org.junit.vintage</groupId>
<artifactId>junit-vintage-engine</artifactId>
<scope>test</scope>
</dependency>
// in Gradle
dependencies {
testCompile 'org.junit.jupiter:junit-jupiter-api:5.3.2'
testCompile 'org.junit.jupiter:junit-jupiter-engine:5.3.2'
testCompile 'org.junit.jupiter:junit-jupiter-params:5.3.2'
testCompile 'org.junit.vintage:junit-vintage-engine:5.3.2'
}

Your old JUnit 4 tests need not be migrated, because JUnit 5 has a test engine, that can run JUnit 4 tests with JUnit 5. So use JUnit 5 for new tests and only migrate JUnit 4 tests if you have to touch them anyway.

Although you can’t update your production code to a newer Java version, it has some benefit to update your test code to a newer one.

The biggest benefit is that you can start learning new language feature during your daily work when you write tests. You don’t make the beginner’s mistake in the production code. You have access to new tools that can help improve your tests. For example, in JUnit 5 it’s more comfortable to write parameterized tests than in JUnit 4. In my experience, developer writes rather parameterized test with JUnit 5 than with JUnit 4 in a situation where parameterized test make sense.

The above described technique also works for other Java version. For example, your production code is on Java 11 and you want to use Java 12 feature in your test code. Another use case for this technique could be learning another JVM language like Groovy, Kotlin or Clojure in your daily work. Then use the new language in your test code.

For Maven projects, this approach has one little pitfall. IntelliJ IDEA ignores the Java version configuration for test code. It uses the configured Java version in production code section for the whole project. An issue is already opened. A workaround is also described in this issue. So only the Maven build gives you the feedback if your production code uses correct Java version. IntelliJ hasn’t this problem for Gradle projects. Here, it uses the Java version just like it is configured in Gradle build file.

The situation in Netbeans looks better for Maven projects. Netbeans loads the Java configuration for Maven project, correctly. For Gradle projects, I couldn’t check it, because in Netbeans 10, there doesn’t yet exist a Gradle plugin (status: January 2019, but for Netbeans 9, so maybe something will come)

Links


Leave a comment

How to Format a Large Code Base Automatically

If you introduce code formatting rules retroactively, you have to solve the problem how to format existing code base according to the new formatting rules. You could checkout every code repository one by one in your IDE and click on Autoformat the whole project. But this is boring and waste of time. Fortunately, Intellij IDEA has a format CLI tool in its installation. You can locate it in the path <your Intellij IDEA installation>/bin. It’s called format.sh.

In the next section I’d like to show you how you can automate formatting big code base. First, I will show the preparation steps like exporting your code formatting rule setting from the IDE. Then, I will demonstrate how to use the CLI-Tool format.sh. At the end, I will show a small Groovy script that query all repositories (in this case they are Git repositories), formatting the code and push it back to the remote SCM.

Preparations

First at all, we need the code formatting rule setting exported from Intellij IDEA. In your Intellij IDEA follow the next step

  1. Open File -> Settings -Editor-> Code Style
  2. Click on Export…
  3. Choose a name for the XML file (for example, Default.xml) and a location where this file should be saved (for example, /home/foo ).

Then, checkout or clone your SCM repository and remember the location where you checkout/clone it (for example, /home/foo/myrepository).

Format Code Base Via format.sh  CLI Tool

Three parameters are important for format.sh:

  • -s : Set a path to Intellij IDEA code style settings .xml file (in our example: /home/foo/Default.xml).
  • -r : Set that directories should be scanned recursively.
  • path<n> : Set a path to a file or a directory that should be formatted (in our example: /home/foo/myrepository).

> ./format.sh
IntelliJ IDEA 2018.2.4, build IC-182.4505.22 Formatter
Usage: format [-h] [-r|-R] [-s|-settings settingsPath] path1 path2...
-h|-help Show a help message and exit.
-s|-settings A path to Intellij IDEA code style settings .xml file.
-r|-R Scan directories recursively.
-m|-mask A comma-separated list of file masks.
path<n> A path to a file or a directory.

> /format.sh -r -s ~/Default.xml ~/myrepository

It’s possible that the tool cancels scanning because of a java.lang.OutOfMemoryError: Java heap space. Then, you have to increase Java’s maximum memory size (-Xmx) in <your Intellij IDEA installation>/bin/idea64.vmoptions.


> nano idea64.vmoptions
-Xms128m
-Xmx750m // <- here increase the maximum memory size
-XX:ReservedCodeCacheSize=240m
-XX:+UseConcMarkSweepGC
-XX:SoftRefLRUPolicyMSPerMB=50
-ea
-Dsun.io.useCanonCaches=false
-Djava.net.preferIPv4Stack=true
-Djdk.http.auth.tunneling.disabledSchemes=""
-XX:+HeapDumpOnOutOfMemoryError
-XX:-OmitStackTraceInFastThrow
-Dawt.useSystemAAFontSettings=lcd
-Dsun.java2d.renderer=sun.java2d.marlin.MarlinRenderingEngine

Groovy Script For Formatting Many Repository In a Row

Now, we want to bring everything together. The script should do four things:

  1. Find all repository URLs whose code has to be formatted.
  2. Check out / Clone the repositories.
  3. Format the code in all branches of each repostory.
  4. Commit and push the change to the remote SCM.

I choose Git as SCM in my example. The finding of the repository URLs depends on the Git Management System (like BitBucket, Gitlab, SCM Manager etc.) that you use. But the approach is in all system the same:

  1. Call the RESTful API of your Git Management System.
  2. Parse the JSON object in the response after the URLs.

For example, in BitBucket it’s like that:



@Grab('org.codehaus.groovy.modules.http-builder:http-builder:0.7.1')
import groovyx.net.http.*

import static groovyx.net.http.ContentType.*
import static groovyx.net.http.Method.*

def cloneUrlsForProject() {

    def projectUrl = "https://scm/rest/api/1.0/projects/PROJECT_KEY/repos?limit=1000"
    def token = "BITBUCKET_TOKEN"

    def projects = []
    def cloneUrls = []

    def http = new HTTPBuilder(projectUrl)
    http.request(GET) {
        headers."Accept" = "application/json"
        headers."Authorization" = "Bearer ${token}"

        response.success = { resp -> projects = new JsonSlurper().parseText(resp.entity.content.text)}

        response.failure = { resp ->
            throw new RuntimeException("Error fetching clone urls for '${projectKey}': ${resp.statusLine}")
        }
    }

    projects.values.each { value ->
        def cloneLink = value.links.clone.find { it.name == "ssh" }
        cloneUrls.add(cloneLink.href)
    }

    return cloneUrls
}

Then, we have to clone the repositories and checkout each branch. In each branch, the format.sh has to be called. For the git operation, we use the jgit library and for the format.sh call we use a Groovy feature for process calling. In Groovy it’s possible to define the command as a String and then to call the method execute() on this String like “ls -l”.execute(). So the Groovy script for the last three tasks would be looked like that:

#!/usr/bin/env groovy
@Grab('org.eclipse.jgit:org.eclipse.jgit:5.1.2.201810061102-r')
import jgit.*
import org.eclipse.jgit.api.CreateBranchCommand
import org.eclipse.jgit.api.Git
import org.eclipse.jgit.api.ListBranchCommand
import org.eclipse.jgit.transport.UsernamePasswordCredentialsProvider

import java.nio.file.Files


def intellijHome = 'path to your idea home folder'
def codeFormatterSetting = 'path to your exported code formatter setting file'
def allRepositoriesUrls = ["http://scm/repo1","http://scm/repo2"] // for simplifying

allRepositoriesUrls.each { repository ->
    def repositoryName = repository.split('/').flatten().findAll { it != null }.last()
    File localPath = Files.createTempDirectory("${repositoryName}-").toFile()
    println "Clone ${repository} to ${localPath}"
    Git.cloneRepository()
       .setURI(repository)
       .setDirectory(localPath)
       .setNoCheckout(true)
       .setCredentialsProvider(new UsernamePasswordCredentialsProvider("user", "password")) // only needed when clone url is https / http
       .call()
       .withCloseable { git ->
        def remoteBranches = git.branchList().setListMode(ListBranchCommand.ListMode.REMOTE).call()
        def remoteBranchNames = remoteBranches.collect { it.name.replace('refs/remotes/origin/', '') }

        println "Found the following branches: ${remoteBranchNames}"

        remoteBranchNames.each { remoteBranch ->
            println "Checkout branch $remoteBranch"
            git.checkout()
               .setName(remoteBranch)
               .setCreateBranch(true)
               .setUpstreamMode(CreateBranchCommand.SetupUpstreamMode.TRACK)
               .setStartPoint("origin/" + remoteBranch)
               .call()

            def formatCommand = "$intellijHome/bin/format.sh -r -s $codeFormatterSetting $localPath"

            println formatCommand.execute().text

            git.add()
               .addFilepattern('.')
               .call()
            git.commit()
               .setAuthor("Automator", "no-reply@yourcompany.com")
               .setMessage('Format code according to IntelliJ setting.')
               .call()

            println "Commit successful!"
        }

        git.push()
           .setCredentialsProvider(new UsernamePasswordCredentialsProvider("user", "password")) // only needed when clone url is https / http
           .setPushAll()
           .call()

        println "Push is done"

    }
}

Do you have another approach? Let me know and write a comment below.


Leave a comment

Running Ansible on a Windows System

On my last conference talk (it was about Ansible and Docker at DevOpsCon in Berlin), I was asked what is the best way to run Ansible on a Windows system. Ansible itself requires a Linux-based system as the control machine. When I have to develop on a Windows machine, I install a Linux-based virtual machine to run the Ansible’s playbooks inside the virtual machine. I set up the virtual machine with Virtualbox and Vagrant. This tools allow me to share the playbooks easily between host and the virtual machine. so I can develop the playbook on the windows system and the virtual machine can have a headless setup. The next section shows you how to set up this tool chain.

 Tool Chain Setup

 At first, install VirtualBox and Vagrant on your machine. I additionally use Babun, a windows shell based on Cygwin and oh-my-zsh, for a better shell experience on Windows, but this isn’t necessary. Then, go to the directory (let’s called it ansible-workspace), where your Ansible’s playbooks are located. Create there a Vagrant configuration file with the command vagrant init:
ansible-workspace
├── inventories
│   ├── production
│   └── test
├── README.md
├── roles
│   ├── deploy-on-tomcat
│   │   ├── defaults
│   │   │   └── main.yml
│   │   └── tasks
│   │       ├── cleanup-webapp.yml
│   │       ├── deploy-webapp.yml
│   │       ├── main.yml
│   │       ├── start-tomcat.yml
│   │       └── stop-tomcat.yml
│   ├── jdk
│   │   └── tasks
│   │       └── main.yml
│   └── tomcat8
│       ├── defaults
│       │   └── main.yml
│       ├── files
│       │   └── init.d
│       │       └── tomcat
│       ├── tasks
│       │   └── main.yml
│       └── templates
│           └── setenv.sh.j2
├── demo-app-ansible-deploy-1.0-SNAPSHOT.war
├── deploy-demo.yml
├── inventories
│   ├── production
│   └── test
├── roles
│   ├── deploy-on-tomcat
│   │   ├── defaults
│   │   │   └── main.yml
│   │   └── tasks
│   │       ├── cleanup-webapp.yml
│   │       ├── deploy-webapp.yml
│   │       ├── main.yml
│   │       ├── start-tomcat.yml
│   │       └── stop-tomcat.yml
│   ├── jdk
│   │   └── tasks
│   │       └── main.yml
│   └── tomcat8
│       ├── defaults
│       │   └── main.yml
│       ├── files
│       │   └── init.d
│       │       └── tomcat
│       ├── tasks
│       │   └── main.yml
│       └── templates
│           └── setenv.sh.j2
├── setup-app-roles.yml
├── setup-app.yml
└── Vagrantfile

├── setup-app-roles.yml
├── setup-app.yml
└── Vagrantfile


Now, we have to choose a so-called Vagrant Box on Vagrant Cloud. A box is the package format for a Vagrant environment. It depends on the provider and the operation system that you choose to use. In our case, it is a Virtualbox VM image based on a minimal Ubuntu 18.04 system (box name is bento/ubuntu-18.04 ). This box will be configured in our Vagrantfile:

Vagrant.configure("2") do |config|
  config.vm.box = "bento/ubuntu-18.04"
end

The next step is to ensure that Ansible will be installed in the box. Thus, we use the shell provisioner of Vagrant. The Vagranfile will be extended by the provisioning code:

Vagrant.configure("2") do |config|
  # ... other Vagrant configuration
  config.vm.provision "shell", inline: &lt;&lt;-SHELL
    sudo apt-get update -y
    sudo apt-get install -y software-properties-common
    sudo apt-add-repository ppa:ansible/ansible
    sudo apt-get update -y
    sudo apt-get install -y ansible
    # ... other Vagrant provision steps
  SHELL
end
end

The last step is to copy the SSH credential into the Vagrant box. Thus, we mark the SSH credential folder of the host system as a Shared folder, so that we can copy them to the SSH config folder inside the box.
Vagrant.configure("2") do |config|
 
  # ... other Vagrant configuration
  config.vm.synced_folder ".", "/vagrant"
  config.vm.synced_folder "path to your ssh config", "/home/vagrant/ssh-host"
  # ... other Vagrant configuration

  config.vm.provision "shell", inline: &lt;&lt;-SHELL
    # ... other Vagrant provision steps
    cp /home/vagrant/ssh-host/* /home/vagrant/.ssh/.
  SHELL
end

On Github’s Gist you can found the whole Vagrantfile.

Workflow

After setting up the tool chain let’s have a look how to work with it. I write my Ansible playbook on the Windows system and run them from the Linux guest system against the remote hosts. For running the Ansible playbooks we have to start the Vagrant box.
> cd ansible-workspace
> vagrant up

When the Vagrant box is ready to use, we can jump into the box with:
 
> vagrant ssh 

You can find the Ansible playbooks inside the box in the folder /vagrant .  In this folder run Ansible:
 
> cd" /vagrant
> ansible-playbook -i inventories/test -u tekkie setup-db.yml

Outlook

Maybe on Windows 10 it’s possible to use Ansible natively, because of the Linux subsystem. But I don’t try it out. Some Docker fans would prefer a container instead of a virtual machine. But remember, before Windows 10 Docker runs on Windows in a virtual machine, so therefore, I don’t see a benefit for using Docker instead of a virtual machine. But of course with Windows 10 native container support a setup with Docker is a good alternative if Ansible doesn’t run on the Linux subsystem.
Do you another idea or approach? Let me know and write a comment.

Links

  1. VirtualBox
  2. Vagrant
  3. Whole Vagrantfile on Github.

 


Leave a comment

Apache2 as Reverse Proxy for NPM Registry Proxies in Sonatype Nexus 3

I use a NPM registry proxy in Sonatype Nexus 3 behind an Apache2 as reverse proxy. With the “standard” Apache2 VirtualHost configuration


<VirtualHost:80>

  ProxyRequests Off
  <Proxy *>
    Order deny,allow
    Allow from all
  </Proxy>
  ProxyPass / http://localhost:8081/
  ProxyPassReverse / http://localhost:8081/

</VirtualHost:80>

I got following failure when I tried to install the dependency @sinonjs/formatio.

$ yarn add @sinonjs/formatio --verbose
yarn add v1.3.2
warning package.json: No license field
verbose 0.337 Checking for configuration file "/home/sparsick/dev/workspace/yarn-test-module/.npmrc".
verbose 0.337 Checking for configuration file "/home/sparsick/.npmrc".
verbose 0.337 Found configuration file "/home/sparsick/.npmrc".
verbose 0.337 Checking for configuration file "/usr/etc/npmrc".
verbose 0.338 Found configuration file "/usr/etc/npmrc".
verbose 0.338 Checking for configuration file "/home/sparsick/dev/workspace/yarn-test-module/.npmrc".
verbose 0.338 Checking for configuration file "/home/sparsick/dev/workspace/.npmrc".
verbose 0.338 Checking for configuration file "/home/sparsick/dev/.npmrc".
verbose 0.338 Checking for configuration file "/home/sparsick/.npmrc".
verbose 0.338 Found configuration file "/home/sparsick/.npmrc".
verbose 0.338 Checking for configuration file "/home/.npmrc".
verbose 0.341 Checking for configuration file "/home/sparsick/dev/workspace/yarn-test-module/.yarnrc".
verbose 0.342 Found configuration file "/home/sparsick/dev/workspace/yarn-test-module/.yarnrc".
verbose 0.343 Checking for configuration file "/home/sparsick/.yarnrc".
verbose 0.344 Found configuration file "/home/sparsick/.yarnrc".
verbose 0.344 Checking for configuration file "/usr/etc/yarnrc".
verbose 0.344 Checking for configuration file "/home/sparsick/dev/workspace/yarn-test-module/.yarnrc".
verbose 0.345 Found configuration file "/home/sparsick/dev/workspace/yarn-test-module/.yarnrc".
verbose 0.345 Checking for configuration file "/home/sparsick/dev/workspace/.yarnrc".
verbose 0.345 Checking for configuration file "/home/sparsick/dev/.yarnrc".
verbose 0.345 Checking for configuration file "/home/sparsick/.yarnrc".
verbose 0.345 Found configuration file "/home/sparsick/.yarnrc".
verbose 0.345 Checking for configuration file "/home/.yarnrc".
verbose 0.347 current time: 2018-02-27T08:04:43.357Z
warning yarn-test-module: No license field
[1/4] Resolving packages...
verbose 0.45 Performing "GET" request to "http://mycompany/repository/npm-public/@sinonjs%2fformatio".
verbose 0.55 Request "http://mycompany/repository/npm-public/@sinonjs%2fformatio" finished with status code 404.
verbose 0.551 Error: Couldn't find package "@sinonjs/formatio" on the "npm" registry.
at /usr/lib/node_modules/yarn/lib/cli.js:49061:15
at Generator.next (<anonymous>)
at step (/usr/lib/node_modules/yarn/lib/cli.js:92:30)
at /usr/lib/node_modules/yarn/lib/cli.js:103:13
at <anonymous>
at process._tickCallback (internal/process/next_tick.js:188:7)
error Couldn't find package "@sinonjs/formatio" on the "npm" registry.
info Visit https://yarnpkg.com/en/docs/cli/add for documentation about this command.</span>

The problem is that Apache2 canonicalizes URLs as default. So I have to configure Apache2 to not canonicalize URLs and additionally, I have to allow encoded slashes:


<VirtualHost:80>

  ProxyRequests Off
  <Proxy *>
    Order deny,allow
    Allow from all
  </Proxy>
  ProxyPass / http://localhost:8081/ nocanon
  ProxyPassReverse / http://localhost:8081/

  AllowEncodedSlashes NoDecode

</VirtualHost:80>

With the above Apache2 Virtualhost configuration I could install my dependency via the NPM registry proxy.


$ yarn add @sinonjs/formatio

yarn add v1.3.2
warning package.json: No license field
info No lockfile found.
warning yarn-test-module: No license field
[1/4] Resolving packages...
[2/4] Fetching packages...
[3/4] Linking dependencies...
[4/4] Building fresh packages...
success Saved lockfile.
success Saved 2 new dependencies.
├─ @sinonjs/formatio@2.0.0
└─ samsam@1.3.0
warning yarn-test-module: No license field
Done in 0.60s.

Big thanks to Sonatype support team, that gave me this advice.