vmtl-adsk / spark-learning

Kafka-Spark jobs orchestrated with Airflow

Spark-Kafka jobs orchestrated with Airflow

The following instruments are used in the project:

• Poetry: as a project dependency management tool
• FastAPI: used to immitate a data stream (a request sent and processed in a Faust-Kafka task)
• Faust: Kafka-Python library
• SQLalchemy: to persist data from FastAPI to a Postgresql database
• Spark: to process a Kafka stream and persist the Data Stream to a Cassandra database
• Airflow: to trigger tasks execution
• Docker: to run everything in containers
• Greate Expectations: to validate data on all the steps of the pipeline

Installation

• Install Docker
• Install all project dependencies spark-learning: poetry install

Airflow

• (Optional) Change Airflow home location (if running Airflow locally) export AIRFLOW_HOME=~spark-learning/airflow
• Create a custom Airflow docker image containing the project and its dependencies spark-learning: docker build . --file airflow/Dockerfile --tag my_airflow_image
• Start Airflow containers spark-learning: dc -f airflow/docker-compose.yaml up
• Install project dependencies on webserver and worker containers:
  • pip install poetry
  • poetry config virtualenvs.create false && poetry install

FastAPI

• Create a FastAPI-Uvicorn image spark-learning/fastapi: docker build . --tag my_fastapi_image
• Start FastAPI container spark-learning/fastapi: docker-compose up

RUN

• Start Docker Kafka, Spark, Cassandra containers spark-learning: docker-compose up
• (optional) docker-compose up --scale spark-worker=3 to spin up the Spark additional spark workers
• Start faust-kafka worker (Required if without Airflow) python data_processing/faust_kafka_stream.py worker -l info
• Start spark (Required if without Airflow) python fastapi/spark.py
• Test run a task in a DAG (Example) airflow tasks test airflow_dag_1 print_date 2015-06-01

pip install poetry && poetry config virtualenvs.create false && poetry install

Helpers

Postgres

• Access postgres container docker exec -it airflow_postgres_1 bash
• Access a DB in the container psql --host=airflow_postgres_1 --dbname=vmtl --username=docker

Cassandra

• Access cassandra container docker exec -it spark-cassandra cqlsh

Kafka

• Add more kafka brokers docker-compose scale kafka=3
• Access a container as a root user docker exec -it -u=0 airflow_airflow-webserver_1 bash

Airflow

• Acecess webserver container docker exec -it airflow_airflow-webserver_1 bash
• Manually start Airflow UI web server airflow webserver --debug &

Great Expectations

• Change default port for Jypiter notebooks in cas of ports conflict export GE_JUPYTER_CMD='jupyter notebook --port=8187'
• Create a datasource great_expectations datasource new
• Create an expectations suite great_expectations suite new
• Create a checkpoint great_expectations checkpoint new