Skip to main content

Developing Locally

Airbyte development is broken into two activities, connector development and platform development. Connector development is largely done in Python by community contributors, though sometimes Java is used for performance reasons. Platform development is done in Java and Kotlin. In addition to the Java and Kotlin code, the Platform also consists of the UI. The UI is developed in TypeScript using React.

Submitting Code

If you would like to submit code to Airbyte, please follow the Pull Request Handbook guide when creating Github Pull Requests. When you are ready to submit code, use the Submit a New Connector document to make sure that the process can go as smoothly as possible.

Prerequisites

info

Manually switching between different language versions can be difficult. We recommend using a version manager such as pyenv or jenv.

The following technologies are required to build Airbyte locally.

  1. Java 21
  2. Node 20
  3. Python 3.10
  4. Docker
  5. Jq

If you are looking to build connectors, you should also follow the installation instructions for airbyte-ci.

Using abctl for Airbyte development

The guides in this document explain how to develop Connectors and the Airbyte Platform with abctl. You should follow the Quickstart instructions to install abctl.

Kubernetes in Docker (kind) is used by abctl to create a local Kubernetes cluster as a docker container. Once the kind cluster has been created, abctl then uses Helm along with the Airbyte Chart to deploy Airbyte. In order to view logs, debug issues, and managed your Airbyte deployment, you should install the kind command line tools, as well as, kubectl for interacting with the Kubernetes cluster.

To configure your kubectl client so that it can communicate with your abctl cluster, run the following:

kind export kubeconfig -n airbyte-abctl

You can then view any issues with the pods (deployed containers) by running:

kubectl get pods -n airbyte-abctl

Which should output something like:

NAME                                                     READY   STATUS      RESTARTS        AGE
airbyte-abctl-airbyte-bootloader 0/1 Completed 0 4h20m
airbyte-abctl-connector-builder-server-55bc78bd-6sxdp 1/1 Running 0 4h20m
airbyte-abctl-cron-b48bccb78-jnz7b 1/1 Running 0 4h20m
airbyte-abctl-pod-sweeper-pod-sweeper-599fd8f56d-kj5t9 1/1 Running 0 4h20m
airbyte-abctl-server-74465db7fd-gk25q 1/1 Running 0 4h20m
airbyte-abctl-temporal-bbb84b56c-jh8x7 1/1 Running 0 4h33m
airbyte-abctl-webapp-745c949464-brpjf 1/1 Running 0 4h20m
airbyte-abctl-worker-79c895c7dc-ssqvc 1/1 Running 0 4h20m
airbyte-db-0 1/1 Running 0 4h34m
airbyte-minio-0 1/1 Running 0 4h34m

Viewing logs for a particular pod can be done by running:

kubectl logs -n airbyte-abctl airbyte-abctl-server-74465db7fd-gk25q

Connector Contributions

  1. Fork the airbyte repository.
  2. Clone the fork on your workstation:
git clone git@github.com:{YOUR_USERNAME}/airbyte.git
cd airbyte
  1. Make sure that abctl is running correctly with the following command abctl local status. Verify the status check was successful

  2. Then, build the connector image:

    • Verify the airbyte-ci tool is installed by running airbyte-ci --help, install the command with the instructions in the Prerequisites if the command is not found.
    • Run airbyte-ci connectors --name source-<source-name> build will build your connector image.
    • Once the command is done, you will find your connector image in your local docker host: airbyte/source-<source-name>:dev.
  3. Verify the image was published locally by running:

docker images ls | grep airbyte/source-<source-name>:dev

You should see output similar to:

airbyte/destination-s3 | dev | 70516a5908ce | 2 minutes ago | 968MB
  1. You can then load the newly built image into the abctl instance using:
kind load docker-image airbyte/source-<source-name>:dev -n airbyte-abctl
info

The above connector image is tagged with dev. You can change this to use another tag if you'd like via the docker image tag command.

  • In your browser, visit http://localhost:8000/
  • Log in
  • Go to Settings (gear icon in lower left corner)
  • Go to Sources or Destinations (depending on which connector you are testing)
  • Update the version number to use your docker image tag (default is dev)
  • Click Change to save the changes

Now when you run a sync with that connector, it will use your local docker image

Connector Specification Caching

The Airbyte Server caches connector specifications for performance reasons. If you update the specification of a connector, you will need to clear this cache so the new changes are registered. To do this:

  • In your browser, visit http://localhost:8000/
  • Log in
  • Go to Settings (gear icon in lower left corner)
  • Go to Sources or Destinations (depending on which connector you are testing)
  • Leave the version set to dev
  • Click Change to save the changes, which will refresh the dev connectors spec

Platform Contributions

  1. Fork the airbyte-platform repository.
  2. Clone the fork on your workstation:
git clone git@github.com:{YOUR_USERNAME}/airbyte-platform.git
cd airbyte-platform
  1. Make sure that abctl is running correctly with the following command abctl local status. Verify the status check was successful

Build with gradle

To compile and build the platform, run the following command in your local airbyte-platform repository:

./gradlew build

This will build all the code and run all the unit tests.

./gradlew build creates all the necessary artifacts (Webapp, Jars, and Docker images) so that you can run Airbyte locally. Since this builds everything, it can take some time.

info

If running tasks on a subproject, you must prepend :oss to the project in gradlew. For example, to build the airbyte-cron project the command would look like: ./gradlew :oss:airbyte-cron:build.

info

Gradle will use all CPU cores by default. If Gradle uses too much/too little CPU, tuning the number of CPU cores it uses to better suit a dev's need can help.

Adjust this by either, 1. Setting an env var: export GRADLE_OPTS="-Dorg.gradle.workers.max=3". 2. Setting a cli option: ./gradlew build --max-workers 3 3. Setting the org.gradle.workers.max property in the gradle.properties file.

A good rule of thumb is to set this to (# of cores - 1).

info

On Mac, if you run into an error while compiling openssl (this happens when running pip install), you may need to explicitly add these flags to your bash profile so that the C compiler can find the appropriate libraries.

export LDFLAGS="-L/usr/local/opt/openssl/lib"
export CPPFLAGS="-I/usr/local/opt/openssl/include"

Using the images in abctl

Once you have successfully built the Platform images, you can load them into the abctl Kind cluster, for example:

kind load docker-image airbyte/server:dev --name airbyte-abctl

Adjust the image for the Airbyte component that you would like to test. Then you can adjust your vaulues.yaml file to use the dev tag for the component, e.g.

server:
image:
tag: dev

Then redeploy abctl by running:

abctl local install --values values.yaml

Run platform acceptance tests

In your local airbyte-platform repository, run the following commands to run acceptance (end-to-end) tests for the platform:

./gradlew clean build
./gradlew :oss:airbyte-tests:acceptanceTests

Test containers start Airbyte locally, run the tests, and shutdown Airbyte after running the tests. If you want to run acceptance tests against local Airbyte that is not managed by the test containers, you need to set USE_EXTERNAL_DEPLOYMENT environment variable to true:

USE_EXTERNAL_DEPLOYMENT=true ./gradlew :oss:airbyte-tests:acceptanceTests

Webapp Contributions

To develop features in the Airbyte Webapp, you must first bring up an instance of Airbyte on TCP port 8001. To do this using abctl, first follow the Quickstart to install abctl. Then run the following:

abctl local install --port 8001

Disabling Authentication

It may be convenient to turn off authentication. If you wish to turn off authentication, create a new text file named: values.yaml and copy the follow into the file:

global:
auth:
enabled: false

Then you can run abctl with the following:

abctl local install --port 8001 --values ./values.yaml

Installing Dependencies

  • Install nvm (Node Version Manager) if not installed
  • Use nvm to install the required node version:
cd airbyte-webapp
nvm install
  • Install the pnpm package manager in the required version. You can use Node's corepack for that:
corepack enable && corepack install

Running the Webapp

  • Start up the react app.
pnpm install
pnpm start

Formatting code

Airbyte runs a code formatter as part of the build to enforce code styles. You should run the formatter yourself before submitting a PR (otherwise the build will fail).

The command to run formatting varies slightly depending on which part of the codebase you are working in.

Connector

We wrapped our code formatting tools in pre-commit. You can install this and other local dev tools by running make tools.install.

You can run pre-commit to format modified files, or pre-commit run --all-files to format all the code your local airbyte repository.

We wrapped this command in a pre-push hook which you can enable with:

make tools.git-hooks.install

You can also uninstall git hooks with:

make tools.git-hooks.clean

Platform

If you are working in the platform run ./gradlew format from the root of the airbyte-platform repository.

Troubleshooting

Resetting the Airbyte developer environment

Sometimes you'll want to reset the data in your local environment. One common case for this is if you are updating an connector's entry in the database (airbyte-config-oss/init-oss/src/main/resources/config), often the easiest thing to do, is wipe the local database and start it from scratch. To reset your data back to a clean install of Airbyte, follow these steps:

  • Make sure you are in your local airbyte-platform repository

  • Delete the datastore volumes in docker

abctl local uninstall --persisted
  • Remove the abctl data on disk
rm -rf ~/.airbyte/
  • Rebuild the project
./gradlew clean build
abctl local install --values values.yaml

If you are working on python connectors, you may also need to reset the virtualenv and re-install the connector's dependencies.

# Assuming you have a virtualenv loaded into your shell
deactivate

# From the connector's directory
# remove the venv directory entirely
rm -rf .venv

# make and activate a new venv
python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"