pipelines/test/kfp-functional-test

kfp-functional-test

Updating python dependencies

pip-tools is used to manage python dependencies. To update dependencies:

1. edit requirements.in
2. run

   pip-compile requirements.in
   

   to update and pin the transitive dependencies.

Run kfp-functional-test in local

Via python

1. run

    gcloud auth application-default login
   

   acquire new user credentials to use for Application Default Credentials.

2. go to the root directory of kubeflow pipelines project, run

   cd {YOUR_ROOT_DIRECTORY_OF_KUBEFLOW_PIPELINES}
   python3 ./test/kfp-functional-test/run_kfp_functional_test.py  --host "https://$(curl https://raw.githubusercontent.com/kubeflow/testing/master/test-infra/kfp/endpoint)"
   

Via docker

1. run

   gcloud auth application-default login
   

   acquire new user credentials to use for Application Default Credentials. Credentials saved to file with {CREDENTIALS_PATH} similar to: [$HOME/.config/gcloud/application_default_credentials.json]

2. copy the Credentials to the temp folder

    cp {CREDENTIALS_PATH} /tmp/keys/{FILENAME}.json
   ```
   

3. Provide authentication credentials by setting the environment variable GOOGLE_APPLICATION_CREDENTIALS. Replace [PATH] with the file path of the JSON file that contains your credentials. run

   export GOOGLE_APPLICATION_CREDENTIALS="/tmp/keys/{FILENAME}.json"
   

4. go to the root directory of kubeflow pipelines project and run

   cd {YOUR_ROOT_DIRECTORY_OF_KUBEFLOW_PIPELINES}
   docker run -it -v $(pwd):/tmp/src -w /tmp/src -e GOOGLE_APPLICATION_CREDENTIALS=/tmp/keys/{FILENAME}.json \
   -v $GOOGLE_APPLICATION_CREDENTIALS:/tmp/keys/{FILENAME}.json:ro \
   python:3.9-slim /tmp/src/test/kfp-functional-test/kfp-functional-test.sh