embedded-elasticsearch

Small utility for creating integration tests that use Elasticsearch. Instead of using Node it downloads Elasticsearch in specified version and starts it in a separate process. It also allows you to install required plugins which is not possible when using NodeBuilder. Utility was tested with 1.x, 2.x and 5.x versions of Elasticsearch.

Introduction

All you need to do to use this tool is create EmbeddedElastic instance. To do so, use provided builder:

final embeddedElastic = EmbeddedElastic.builder()
        .withElasticVersion("5.0.0")
        .withSetting(PopularProperties.TRANSPORT_TCP_PORT, 9350)
        .withSetting(PopularProperties.CLUSTER_NAME, "my_cluster")
        .withPlugin("analysis-stempel")
        .withIndex("cars", IndexSettings.builder()
            .withType("car", getSystemResourceAsStream("car-mapping.json"))
            .build())
        .withIndex("books", IndexSettings.builder()
            .withType(PAPER_BOOK_INDEX_TYPE, getSystemResourceAsStream("paper-book-mapping.json"))
            .withType("audio_book", getSystemResourceAsStream("audio-book-mapping.json"))
            .withSettings(getSystemResourceAsStream("elastic-settings.json"))
            .build())
        .build()
        .start()

When you are done with creating it, you can start it real simple:

embeddedElastic.start()

And that's all, you can connect to your embedded-elastic instance on specified port and use it in your tests.

Available builder options

Method	Description
withElasticVersion(String version)	version of Elasticsearch; based on that version download url to official Elasticsearch repository will be created
withDownloadUrl(URL downloadUrl)	if you prefer to download Elasticsearch from a different location than official repositories you can do that using this method
withSetting(String key, Object value)	setting name and value as in elasticsearch.yml file
withPlugin(String expression)	plugin that should be installed into Elasticsearch; treat expression as argument to ./elasticsearch-plugin install <expression> command; use multiple times for multiple plugins
withIndex(String indexName, IndexSettings indexSettings)	specify index that should be created and managed by EmbeddedElastic
withStartTimeout(long value, TimeUnit unit)	specify timeout you give Elasticsearch to start
withInstallationDirectory(File installationDirectory)	specify custom installation directory
withDownloadDirectory(File downloadDirectory)	specify custom download directory where downloaded distribution packages will be saved
withCleanInstallationDirectoryOnStop(boolean cleanInstallationDirectoryOnStop)	specify whether clean the installation directory after Elasticsearch stop
withEsJavaOpts(String javaOpts)	value of ES_JAVA_OPTS variable to be set for Elasticsearch process
getTransportTcpPort()	get transport tcp port number used by Elasticsearch instance
getHttpPort()	get http port number used by Elasticsearch instance

Available IndexSettings.Builder options

Method	Description
withType(String type, String mapping)	specify type and it's mappings
withSettings(String settings)	specify index settings

Available operations

EmbeddedElastic provides following operations:

Method	Description
start()	downloads Elasticsearch and specified plugins, setups everything and finally starts your Elasticsearch instance
stop()	stops your Elasticsearch instance and removes all data
index	index your document, comes with variants that take only document, or document and it's id
deleteIndex(String indexName), deleteIndices()	deletes index with name specified during EmbeddedElastic creation
createIndex(String indexName), createIndices()	creates index with name specified during EmbeddedElastic creation; note that this index is created during EmbeddedElastic startup, you will need this method only if you deleted your index using deleteIndex method
recreateIndex(String indexName), recreateIndices()	combination of deleteIndex and createIndex
refreshIndices()	refresh index; useful when you make changes in different thread, and want to check results instantly in tests

Example

If you want to see example, look at this spec: pl.allegro.tech.search.embeddedelasticsearch.EmbeddedElasticSpec

Dependency

To start using embedded-elasticsearch in your project add it as a test dependency:

Gradle:

testCompile 'pl.allegro.tech:embedded-elasticsearch:2.1.0'

Maven:

<dependency>
    <groupId>pl.allegro.tech</groupId>
    <artifactId>embedded-elasticsearch</artifactId>
    <version>2.1.0</version>
    <scope>testCompile</scope>
</dependency>

SBT:

libraryDependencies ++= Seq("pl.allegro.tech" % "embedded-elasticsearch" % "2.1.0" % "test")

Known problems

If you build your project on Travis, you may have problems with OOM errors when using default settings. You can change Elasticsearch memory settings using withEsJavaOpts method. Example (from spec pl.allegro.tech.embeddedelasticsearch.EmbeddedElasticSpec):

    static EmbeddedElastic embeddedElastic = EmbeddedElastic.builder()
            .withElasticVersion(ELASTIC_VERSION)
            .withSetting(TRANSPORT_TCP_PORT, TRANSPORT_TCP_PORT_VALUE)
            .withSetting(CLUSTER_NAME, CLUSTER_NAME_VALUE)
            .withEsJavaOpts("-Xms128m -Xmx512m")
            .withIndex(CARS_INDEX_NAME, CARS_INDEX)
            .withIndex(BOOKS_INDEX_NAME, BOOKS_INDEX)
            .withStartTimeout(1, MINUTES)
            .build()
            .start()

Running more then one Elasticsearch instance

There are cases where you might want to run more then one Elasticsearch instance e.g.:

• running tests of one project in parallel (e.g. using gradle --parallel or mvn -T1C)
• running tests of different projects on the same physical mashine (e.g. Jenkins jobs running on the same server)
• running integration tests wich require more then elastic search instance

In such situations you should use distinct values for following settings for each instance:

• withSetting(PopularProperties.TRANSPORT_TCP_PORT, ...)
• withSetting(PopularProperties.HTTP_PORT, ...)
• withInstallationDirectory(...)

With such configuration embedded-elasticsearch will redonload elasticsearch instllation package for every distinct instalation directory. To avoid whis behavior and thus reuse downloaded installation package you should set common location of downloaded files with withDownloadDirectory(...) for every embedded-elasticsearch configuration.

License

embedded-elasticsearch is published under Apache License 2.0.