Kedro Airflow Using Docker

Create a kedro project and run kedro pipelines on Airflow platform
Pre-process data, train model, evaluate and predict

Prerequisites

• Docker Compose
• Airflow 1.10.9
• Kedro 0.16.6
• Kedro-Airflow 0.3.0
• scikit-learn 0.23.0
• pickle 0.0.11

Workflows

• Read data from csv files and excel file as well, pre-process files and then save csv files
• Split data and then save pickle files
• Read pickle files, run train model and also save the regression model(pickle format)
• Load the regression model and run Predict from the pickle model

Build

• Build image

  ./scripts/build.sh
  

• Run Airflow Webserver

  docker-compose up
  

• Run Airflow Scheduler

  docker exec -it kedro-airflow-docker_webserver_1 bash
  airflow scheduler
  

• Open web browser and click trigger dag

  http://localhost:8080
  

• Open Visualise Pipelines

  docker exec -it kedro-airflow-docker_webserver_1 bash
  kedro viz --host 0.0.0.0
  

  Open the browser with http://localhost:4141

Issues

• NOT Support for MemoryDataSet

  {{api.py:296}} INFO - Loading: /usr/local/airflow/example/conf/base/logging.yml
  /usr/local/lib/python3.7/site-packages/fsspec/implementations/local.py:33: FutureWarning: The default value of auto_mkdir=True has been deprecated and will be changed to auto_mkdir=False by default in a future release.
  FutureWarning,
  2020-12-23 03:17:14,387 - airflow.models.dagbag.DagBag - ERROR - Failed to import: /usr/local/airflow/dags/example_dag.py
  Traceback (most recent call last):
  File "/usr/local/lib/python3.7/site-packages/airflow/models/dagbag.py", line 243, in process_file
      m = imp.load_source(mod_name, filepath)
  File "/usr/local/lib/python3.7/imp.py", line 171, in load_source
      module = _load(spec)
  File "<frozen importlib._bootstrap>", line 696, in _load
  File "<frozen importlib._bootstrap>", line 677, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 728, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "/usr/local/airflow/dags/example_dag.py", line 95, in <module>
      runner.run(pipeline, data_catalog)
  File "/usr/local/lib/python3.7/site-packages/kedro_airflow/runner.py", line 115, in run
      catalog.add(ds_name, self.create_default_data_set(ds_name))
  File "/usr/local/lib/python3.7/site-packages/kedro_airflow/runner.py", line 90, in create_default_data_set
      "AirflowRunner does not support unregistered data sets.".format(ds_name)
  ValueError: Data set 'example_train_y' is not registered in the data catalog.
  AirflowRunner does not support unregistered data sets.
  
  

  Fixed: define output data files for each node in catalog.yml(save output files for each node)

• Could not load Excel Data Set

  kedro.io.core.DataSetError: Failed while loading data from data set ExcelDataSet
  (filepath=/home/kedro/data/01_raw/shuttles.xlsx,
  load_args={'engine': xlrd}, protocol=file, save_args={'index': False},
  writer_args={'engine': xlsxwriter}).
  Excel xlsx file; not supported
  

  Fixed: xlrd==1.2.0

Results

• Kedro Visualise Pipelines

• Kedro Logs

• Airflow Dashboard

• Airflow Logs

Reference

• Create a Pipeline
• Kedro Airflow Test