- Developer Guide
So you want to contribute code to OpenSearch? Excellent! We're glad you're here. Here's what you need to do.
Fork opensearch-project/OpenSearch and clone locally, e.g. git clone https://github.com/[your username]/OpenSearch.git.
OpenSearch builds using Java 11 at a minimum, using the Adoptium distribution. This means you must have a JDK 11 installed with the environment variable JAVA_HOME referencing the path to Java home for your JDK 11 installation, e.g. JAVA_HOME=/usr/lib/jvm/jdk-11. This is configured in buildSrc/build.gradle and distribution/tools/java-version-checker/build.gradle.
allprojects {
targetCompatibility = JavaVersion.VERSION_11
sourceCompatibility = JavaVersion.VERSION_11
}
sourceCompatibility = JavaVersion.VERSION_11
targetCompatibility = JavaVersion.VERSION_11
Download Java 11 from here.
To run the full suite of tests, download and install JDK 14 and set JAVA11_HOME, and JAVA14_HOME. They are required by the backwards compatibility test.
By default, the test tasks use bundled JDK runtime, configured in buildSrc/version.properties, and set to JDK 17 (LTS).
bundled_jdk_vendor = adoptium
bundled_jdk = 17.0.2+8
Other kind of test tasks (integration, cluster, etc.) use the same runtime as JAVA_HOME. However, the build also supports compiling with one version of JDK, and testing on a different version. To do this, set RUNTIME_JAVA_HOME pointing to the Java home of another JDK installation, e.g. RUNTIME_JAVA_HOME=/usr/lib/jvm/jdk-14. Alternatively, the runtime JDK version could be provided as the command line argument, using combination of runtime.java=<major JDK version> property and JAVA<major JDK version>_HOME environment variable, for example ./gradlew -Druntime.java=17 ... (in this case, the tooling expects JAVA17_HOME environment variable to be set).
On Windows, set _JAVA_OPTIONS: -Xmx4096M. You may also need to set LongPathsEnabled=0x1 under Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem.
Download and install Docker, required for building OpenSearch artifacts, and executing certain test suites.
On Windows, use Docker Desktop 3.6. See OpenSearch#1425 for workarounds and issues with Docker Desktop 4.1.1.
To build all distributions of OpenSearch, run:
./gradlew assemble
To build a distribution to run on your local platform, run:
./gradlew localDistro
All distributions built will be under distributions/archives.
OpenSearch uses a Gradle wrapper for its build. Run gradlew on Unix systems, or gradlew.bat on Windows in the root of the repository.
Start by running the test suite with gradlew check. This should complete without errors.
./gradlew check
=======================================
OpenSearch Build Hamster says Hello!
Gradle Version : 6.6.1
OS Info : Linux 5.4.0-1037-aws (amd64)
JDK Version : 11 (JDK)
JAVA_HOME : /usr/lib/jvm/java-11-openjdk-amd64
=======================================
...
BUILD SUCCESSFUL in 14m 50s
2587 actionable tasks: 2450 executed, 137 up-to-date
If the full test suite fails you may want to start with a smaller set.
./gradlew precommit
Run OpenSearch using gradlew run.
./gradlew run
That will build OpenSearch and start it, writing its log above Gradle's status message. We log a lot of stuff on startup, specifically these lines tell you that OpenSearch is ready.
[2020-05-29T14:50:35,167][INFO ][o.e.h.AbstractHttpServerTransport] [runTask-0] publish_address {127.0.0.1:9200}, bound_addresses {[::1]:9200}, {127.0.0.1:9200}
[2020-05-29T14:50:35,169][INFO ][o.e.n.Node ] [runTask-0] started
It's typically easier to wait until the console stops scrolling, and then run curl in another window to check if OpenSearch instance is running.
curl localhost:9200
{
"name" : "runTask-0",
"cluster_name" : "runTask",
"cluster_uuid" : "oX_S6cxGSgOr_mNnUxO6yQ",
"version" : {
"number" : "1.0.0-SNAPSHOT",
"build_type" : "tar",
"build_hash" : "0ba0e7cc26060f964fcbf6ee45bae53b3a9941d0",
"build_date" : "2021-04-16T19:45:44.248303Z",
"build_snapshot" : true,
"lucene_version" : "8.7.0",
"minimum_wire_compatibility_version" : "6.8.0",
"minimum_index_compatibility_version" : "6.0.0-beta1"
}
}Use -Dtests.opensearch. to pass additional settings to the running instance. For example, to enable OpenSearch to listen on an external IP address pass -Dtests.opensearch.http.host. Make sure your firewall or security policy allows external connections for this to work.
./gradlew run -Dtests.opensearch.http.host=0.0.0.0When importing into IntelliJ you will need to define an appropriate JDK. The convention is that this SDK should be named "11", and the project import will detect it automatically. For more details on defining an SDK in IntelliJ please refer to this documentation. Note that SDK definitions are global, so you can add the JDK from any project, or after project import. Importing with a missing JDK will still work, IntelliJ will report a problem and will refuse to build until resolved.
You can import the OpenSearch project into IntelliJ IDEA as follows.
- Select File > Open
- In the subsequent dialog navigate to the root
build.gradlefile - In the subsequent dialog select Open as Project
JetBrains Gateway enables development, testing and debugging on remote machines like development servers.
- On the local development machine, download and install the latest thin client from the JetBrains Gateway page.
- Create a new connection to the remote server and install an IntelliJ server support using these instructions.
Follow the IntelliJ IDEA instructions post a successful connection.
Follow links in the Java Tutorial to install the coding pack and extensions for Java, Gradle tasks, etc. Open the source code directory.
When importing to Eclipse, you need to have Eclipse Buildship plugin installed and, preferably, have JDK 11 set as default JRE in Preferences -> Java -> Installed JREs. Once this is done, generate Eclipse projects using Gradle wrapper:
./gradlew eclipse
You can now import the OpenSearch project into Eclipse as follows.
- Select File > Import -> Existing Gradle Project
- In the subsequent dialog navigate to the root of
build.gradlefile - In the subsequent dialog, if JDK 11 is not set as default JRE, please make sure to check [Override workspace settings], keep [Gradle Wrapper] and provide the correct path to JDK11 using [Java Home] property under [Advanced Options]. Otherwise, you may run into cryptic import failures and only top level project is going to be imported.
- In the subsequent dialog, you should see [Gradle project structure] populated, please click [Finish] to complete the import
Note: it may look non-intuitive why one needs to use Gradle wrapper and then import existing Gradle project (in general, File > Import -> Existing Gradle Project should be enough). Practically, as it stands now, Eclipse Buildship plugin does not import OpenSearch project dependencies correctly but does work in conjunction with Gradle wrapper.
This repository is split into many top level directories. The most important ones are:
Builds our tar and zip archives and our rpm and deb packages.
Libraries used to build other parts of the project. These are meant to be internal rather than general purpose. We have no plans to semver their APIs or accept feature requests for them. We publish them to maven central because they are dependencies of our plugin test framework, high level rest client, and jdbc driver but they really aren't general purpose enough to belong in maven central. We're still working out what to do here.
Features that are shipped with OpenSearch by default but are not built in to the server. We typically separate features from the server because they require permissions that we don't believe all of OpenSearch should have or because they depend on libraries that we don't believe all of OpenSearch should depend on.
For example, reindex requires the connect permission so it can perform reindex-from-remote but we don't believe that the all of OpenSearch should have the "connect". For another example, Painless is implemented using antlr4 and asm and we don't believe that all of OpenSearch should have access to them.
OpenSearch plugins. We decide that a feature should be a plugin rather than shipped as a module because we feel that it is only important to a subset of users, especially if it requires extra dependencies.
The canonical example of this is the ICU analysis plugin. It is important for folks who want the fairly language neutral ICU analyzer but the library to implement the analyzer is 11MB so we don't ship it with OpenSearch by default.
Another example is the discovery-gce plugin. It is vital to folks running in GCP but useless otherwise and it depends on a dozen extra jars.
This is where the community can add experimental features in to OpenSearch. There are three directories inside the sandbox - libs, modules and plugins - which mirror the subdirectories in the project root and have the same guidelines for deciding on where a new feature goes. The artifacts from libs and modules will be automatically included in the snapshot distributions. Once a certain feature is deemed worthy to be included in the OpenSearch release, it will be promoted to the corresponding subdirectory in the project root. Note: The sandbox code do not have any other guarantees such as backwards compatibility or long term support and can be removed at any time.
To exclude the modules from snapshot distributions, use the sandbox.enabled system property.
./gradlew assemble -Dsandbox.enabled=false
Honestly this is kind of in flux and we're not 100% sure where we'll end up. We welcome your thoughts and help.
Right now the directory contains the following.
- Tests that require multiple modules or plugins to work.
- Tests that form a cluster made up of multiple versions of OpenSearch like full cluster restart, rolling restarts, and mixed version tests.
- Tests that test the OpenSearch clients in "interesting" places like the
wildflyproject. - Tests that test OpenSearch in funny configurations like with ingest disabled.
- Tests that need to do strange things like install plugins that thrown uncaught
Throwables or add a shutdown hook.
But we're not convinced that all of these things belong in the qa directory. We're fairly sure that tests that require multiple modules or plugins to work should just pick a "home" plugin. We're fairly sure that the multi-version tests do belong in qa. Beyond that, we're not sure. If you want to add a new qa project, open a PR and be ready to discuss options.
The server component of OpenSearch that contains all of the modules and plugins. Right now things like the high level rest client depend on the server but we'd like to fix that in the future.
Our test framework and test fixtures. We use the test framework for testing the server, the plugins, and modules, and pretty much everything else. We publish the test framework so folks who develop OpenSearch plugins can use it to test the plugins. The test fixtures are external processes that we start before running specific tests that rely on them.
For example, we have an hdfs test that uses mini-hdfs to test our repository-hdfs plugin.
Java files in the OpenSearch codebase are formatted with the Eclipse JDT formatter, using the Spotless Gradle plugin. This plugin is configured on a project-by-project basis, via build.gradle in the root of the repository. So long as at least one project is configured, the formatting check can be run explicitly with:
./gradlew spotlessJavaCheck
The code can be formatted with:
./gradlew spotlessApply
These tasks can also be run for specific subprojects, e.g.
./gradlew server:spotlessJavaCheck
Please follow these formatting guidelines:
- Java indent is 4 spaces
- Line width is 140 characters
- Lines of code surrounded by
// tag::NAMEand// end::NAMEcomments are included in the documentation and should only be 76 characters wide not counting leading indentation. Such regions of code are not formatted automatically as it is not possible to change the line length rule of the formatter for part of a file. Please format such sections sympathetically with the rest of the code, while keeping lines to maximum length of 76 characters. - Wildcard imports (
import foo.bar.baz.*) are forbidden and will cause the build to fail. - If absolutely necessary, you can disable formatting for regions of code with the
// tag::NAMEand// end::NAMEdirectives, but note that these are intended for use in documentation, so please make it clear what you have done, and only do this where the benefit clearly outweighs the decrease in consistency. - Note that JavaDoc and block comments i.e.
/* ... */are not formatted, but line comments i.e// ...are. - There is an implicit rule that negative boolean expressions should use the form
foo == falseinstead of!foofor better readability of the code. While this isn't strictly enforced, if might get called out in PR reviews as something to change.
IntelliJ IDEs can import the settings file, and / or use the Eclipse Code Formatter plugin.
You can also tell Spotless to format a specific file from the command line.
Sometimes Spotless will report a "misbehaving rule which can't make up its mind" and will recommend enabling the paddedCell() setting. If you enabled this settings and run the format check again, Spotless will write files to $PROJECT/build/spotless-diagnose-java/ to aid diagnosis. It writes different copies of the formatted files, so that you can see how they
differ and infer what is the problem.
The paddedCell() option is disabled for normal operation in order to detect any misbehaviour. You can enable the option from the command line by running Gradle with -Dspotless.paddedcell.
Note: if you have imported the project into IntelliJ IDEA the project will be automatically configured to add the correct license header to new source files based on the source location.
We use Gradle to build OpenSearch because it is flexible enough to not only build and package OpenSearch, but also orchestrate all of the ways that we have to test OpenSearch.
Gradle organizes dependencies and build artifacts into "configurations" and allows you to use these configurations arbitrarily. Here are some of the most common configurations in our build and how we use them:
Dependencies that are used by the project at compile and runtime but are not exposed as a compile dependency to other dependent projects. Dependencies added to the implementation configuration are considered an implementation detail that can be changed at a later date without affecting any dependent projects.
Dependencies that are used as compile and runtime dependencies of a project and are considered part of the external api of the project.
Dependencies that not on the classpath at compile time but are on the classpath at runtime. We mostly use this configuration to make sure that we do not accidentally compile against dependencies of our dependencies also known as "transitive" dependencies".
Code that is on the classpath at compile time but that should not be shipped with the project because it is "provided" by the runtime somehow. OpenSearch plugins use this configuration to include dependencies that are bundled with OpenSearch's server.
Code that is on the classpath for compiling tests that are part of this project but not production code. The canonical example
of this is junit.
The Distribution Download plugin downloads the latest version of OpenSearch by default, and supports overriding this behavior by setting customDistributionUrl.
./gradlew integTest -PcustomDistributionUrl="https://ci.opensearch.org/ci/dbc/bundle-build/1.2.0/1127/linux/x64/dist/opensearch-1.2.0-linux-x64.tar.gz"
A fat-JAR (or an uber-JAR) is the JAR, which contains classes from all the libraries, on which your project depends and, of course, the classes of current project.
There might be cases where a developer would like to add some custom logic to the code of a module (or multiple modules) and generate a fat-JAR that can be directly used by the dependency management tool. For example, in #3665 a developer wanted to provide a tentative patch as a fat-JAR to a consumer for changes made in the high level REST client.
Use Gradle Shadow plugin.
Add the following to the build.gradle file of the module for which you want to create the fat-JAR, e.g. client/rest-high-level/build.gradle:
apply plugin: 'com.github.johnrengelman.shadow'
Run the shadowJar command using:
./gradlew :client:rest-high-level:shadowJar
This will generate a fat-JAR in the build/distributions folder of the module, e.g. ./client/rest-high-level/build/distributions/opensearch-rest-high-level-client-1.4.0-SNAPSHOT.jar.
You can further customize your fat-JAR by customising the plugin, More information about shadow plugin can be found here.
To use the generated JAR, install the JAR locally, e.g.
mvn install:install-file -Dfile=src/main/resources/opensearch-rest-high-level-client-1.4.0-SNAPSHOT.jar -DgroupId=org.opensearch.client -DartifactId=opensearch-rest-high-level-client -Dversion=1.4.0-SNAPSHOT -Dpackaging=jar -DgeneratePom=true
Refer the installed JAR as any other maven artifact, e.g.
<dependency>
<groupId>org.opensearch.client</groupId>
<artifactId>opensearch-rest-high-level-client</artifactId>
<version>1.4.0-SNAPSHOT</version>
</dependency>
As you work in the OpenSearch repo you may notice issues getting labeled with component labels. It's a housekeeping task to help group together similar pieces of work. You can pretty much ignore it, but if you're curious, here's what the different labels mean:
Tasks to make sure the build tasks are useful and packaging and distribution are easy.
Includes:
- Gradle for the Core tasks
- Groovy scripts
- build-tools
- Versioning interfaces
- Compatibility
- Javadoc enforcement
APIs and communication mechanisms for external connections to OpenSearch. This includes the “library” directory in OpenSearch (a set of common functions).
Includes:
- Transport layer
- High Level and low level Rest Client
- CLI
Anything touching the plugin infrastructure within core OpenSearch.
Includes:
- API
- SPI
- Plugin interfaces
The critical path of indexing and search, including: Measure index and search, performance, Improving the performance of indexing and search, ensure synchronization OpenSearch APIs with upstream Lucene change (e.g. new field types, changing doc values and codex).
Includes:
- Lucene Structures
- FieldMappers
- QueryBuilders
- DocValues
Making sure OpenSearch can be used as a compute engine.
Includes:
- APIs (suggest supporting a formal API)
- Framework
Work to make sure that OpenSearch can scale in a distributed manner.
Includes:
- Nodes (Cluster Manager, Data, Compute, Ingest, Discovery, etc.)
- Replication & Merge Policies (Document, Segment level)
- Snapshot/Restore (repositories; S3, Azure, GCP, NFS)
- Translog (e.g., OpenSearch, Kafka, Kinesis)
- Shard Strategies
- Circuit Breakers
Security is our top priority. Avoid checking in credentials.
Install awslabs/git-secrets by running the following commands.
git clone https://github.com/awslabs/git-secrets.git
cd git-secrets
make install
You can configure git secrets per repository, you need to change the directory to the root of the repository and run the following command.
git secrets --install
✓ Installed commit-msg hook to .git/hooks/commit-msg
✓ Installed pre-commit hook to .git/hooks/pre-commit
✓ Installed prepare-commit-msg hook to .git/hooks/prepare-commit-msg
Then, you need to apply patterns for git-secrets, you can install the AWS standard patterns by running the following command.
git secrets --register-aws
See CONTRIBUTING.
OpenSearch strives for a smooth and easy upgrade experience that is resilient to data loss and corruption while minimizing downtime and ensuring integration with external systems does not unexpectedly break.
To provide these guarantees each version must be designed and developed with forward compatibility in mind. OpenSearch addresses backward and forward compatibility at three different levels: 1. Data, 2. Developer API, 3. User API. These levels and the developer mechanisms to ensure backwards compatibility are provided below.
The data level consists of index and application data file formats. OpenSearch guarantees file formats and indexes are compatible only back to the first release of
the previous major version. If on disk formats or encodings need to be changed (including index data, cluster state, or any other persisted data) developers must
use Version checks accordingly (e.g., Version.onOrAfter, Version.before) to guarantee backwards compatibility.
The Developer API consists of interfaces and foundation software implementations that enable external users to develop new OpenSearch features. This includes
obvious components such as the Plugin framework and less obvious components such as REST Action Handlers. When developing a new feature of OpenSearch it is important
to explicitly mark which implementation components may, or may not, be extended by external implementations. For example, all new API classes with @opensearch.api
signal that the new component may be extended by an external implementation and therefore provide backwards compatibility guarantees. Similarly, any class explicitly
marked with the @opensearch.internal annotation, or not explicitly marked by an annotation should not be extended by external implementation components as it does not
guarantee backwards compatibility and may change at any time. The @deprecated annotation should also be added to any @opensearch.api classes or methods that are
either changed or planned to be removed across minor versions.
The User API consists of integration specifications (e.g., Query Domain Specific Language,
field mappings) and endpoints (e.g., _search,
_cat) users rely on to integrate and use OpenSearch. Backwards compatibility is critical to the
User API, therefore OpenSearch commits to using semantic versioning for all User facing APIs. To support this
developers must leverage Version checks for any user facing endpoints or API specifications that change across minor versions. Developers must also inform
users of any changes by adding the >breaking label on Pull Requests, adding an entry to the CHANGELOG
and a log message to the OpenSearch deprecation log files using the DeprecationLogger.
Rapidly developing new features often benefit from several release cycles before committing to an official and long term supported (LTS) API. To enable this cycle OpenSearch
uses an Experimental Development process leveraging Feature Flags. This allows a feature to be developed using the same process as
a LTS feature but with additional guard rails and communication mechanisms to signal to the users and development community the feature is not yet stable, may change in a future
release, or be removed altogether. Any Developer or User APIs implemented along with the experimental feature should be marked with the @opensearch.experimental annotation to
signal the implementation is not subject to LTS and does not follow backwards compatibility guidelines.
The Github workflow in backport.yml creates backport PRs automatically when the original PR with an appropriate label backport <backport-branch-name> is merged to main with the backport workflow run successfully on the PR. For example, if a PR on main needs to be backported to 1.x branch, add a label backport 1.x to the PR and make sure the backport workflow runs on the PR along with other checks. Once this PR is merged to main, the workflow will create a backport PR to the 1.x branch.
A linter in code-hygiene.yml that validates simple newline and whitespace rules in all sorts of files. It can:
- Recursively check a directory tree for files that do not end in a newline
- Automatically fix these files by adding a newline or trimming extra newlines.
Rules are defined in .linelint.yml.
Executing the binary will automatically search the local directory tree for linting errors.
linelint .
Pass a list of files or directories to limit your search.
linelint README.md LICENSE
The Github workflow in lucene-snapshots.yml is a Github worfklow executable by maintainers to build a top-down snapshot build of lucene.
These snapshots are available to test compatibility with upcoming changes to Lucene by updating the version at version.properties with the version-snapshot-sha version. Example: lucene = 10.0.0-snapshot-2e941fc.
OpenSearch has a very large test suite with long running, often failing (flaky), integration tests. Such individual tests are labelled as Flaky Random Test Failure. Your help is wanted fixing these!
If you encounter a build/test failure in CI that is unrelated to the change in your pull request, it may be a known flaky test, or a new test failure.
- Follow failed CI links, and locate the failing test(s).
- Copy-paste the failure into a comment of your PR.
- Search through issues using the name of the failed test for whether this is a known flaky test.
- If an existing issue is found, paste a link to the known issue in a comment to your PR.
- If no existing issue is found, open one.
- Retry CI via the GitHub UX or by pushing an update to your PR.