Python tool for deploying Airflow Multi-Node Cluster (a.k.a. Celery Executor Setup).
- Python >=3.6 (tested)
pip install airflow-run
To provide a quick way to setup Airflow Multi-Node Cluster (a.k.a. Celery Executor Setup).
- Docker
- Generate config yaml file.
- Run commands to start webserver, scheduler, worker, (rabbitmq, postgres).
- Add dag files and run initdb.
Run the following and follow the prompt to generate config file.
afr --generate_config
afr --run postgresql
afr --run initdb
afr --run rabbitmq
afr --run webserver
afr --run scheduler
afr --run worker --queue {queue name}
afr --run flower
afr --run postgresql --config /my_path/config.yaml
export AIRFLOWRUN_CONFIG_PATH="/some_path/config.yaml"
After running webserver, scheduler and worker (postgres and rabbitmq if needed local instances), Add your dag files in the dags subdirectory in the directory you defined in the config file. (* note: make sure you have the correct user permission in the dags, logs subdirectories.)
That is it!!
private_registry: False
registry_url: registry.hub.docker.com
username: ""
password: ""
repository: pkuong/airflow-run
image: airflow-run
tag: latest
local_dir: {local directory where you want to mount /dags and /logs folder}
webserver_port: 8000
flower_port: 5555
custom_mount_volumes: []
env:
AIRFLOW__CORE__EXECUTOR: CeleryExecutor
AIRFLOW__CORE__LOAD_EXAMPLES: "False"
AIRFLOW__CORE__DAGS_FOLDER: /usr/local/airflow/airflow/dags
AIRFLOW__CORE__LOGS_FOLDER: /usr/local/airflow/airflow/logs
AIRFLOW_HOME: /usr/local/airflow
AIRFLOW__CORE__FERNET_KEY: ""
rabbitmq:
name: rabbitmq
username: {username}
password: {password}
host: {IP}
virtual_host: /
image: rabbitmq:3-management
home: /var/lib/rabbitmq
ui_port: 15672
port: 5672
env:
RABBITMQ_DEFAULT_USER: {username}
RABBITMQ_DEFAULT_PASS: {password}
postgresql:
name: postgresql
username: {username}
password: {password}
host: {host}
image: postgres
data: /var/lib/postgresql/data
port: 5432
env:
PGDATA: /var/lib/postgresql/data/pgdata
POSTGRES_USER: {username}
POSTGRES_PASSWORD: {password}
You can specify custom mount volumes in the container, for example:
custom_mount_volumes:
- host_path: /Users/bob/.aws
container_path: /usr/local/airflow/.aws
This tool is using the following public docker image by default.
https://hub.docker.com/repository/docker/pkuong/airflow-run
If you want to build your own image, you can run the following:
afr --build --config={absolute path to config.yaml} --dockerfile={absolute path to directory which contains Dockerfile}- Paulo Kuong (@pkuong)