federto / Flask-Redis-API-ML-Resnet-50

Flask ML API

Install and run

To run the services using compose:

$ docker-compose up --build -d

To stop the services:

$ docker-compose down

Tests

Integration end-to-end

You must have the full pipeline running (see previous section) and the following Python libraries installed:

• requests==2.28.1
• pytest==7.1.2

You can install them using the file tests/requirements.txt with:

$ pip3 install -r tests/requirements.txt

Then, from the project root folder run:

$ python tests/test_integration.py

If the output looks like this, then the integration tests are passing:

.
----------------------------------------------------------------------
Ran 2 tests in 0.299s

OK

Modules

We make use of multi-stage docker builds so we can have into the same Dockerfile environments for testing and also for deploying our service.

Api

Run:

$ cd api/
$ docker build -t flask_api_test --progress=plain --target test .

You will only need to pay attention to the logs corresponding to the testing code which will look like this:

#10 [test 1/1] RUN ["pytest", "-v", "/src/tests"]
#10 sha256:707efc0d59d04744766193fe6873d212afc0f8e4b28d035a2d2e94b40826604f
#10 0.537 ============================= test session starts ==============================
#10 0.537 platform linux -- Python 3.8.13, pytest-7.1.1, pluggy-1.0.0 -- /usr/local/bin/python
#10 0.537 cachedir: .pytest_cache
#10 0.537 rootdir: /src
#10 0.537 collecting ... collected 4 items
#10 0.748
#10 0.748 tests/test_api.py::TestIntegration::test_bad_parameters PASSED           [ 25%]
#10 0.757 tests/test_api.py::TestEnpointsAvailability::test_feedback PASSED        [ 50%]
#10 0.769 tests/test_api.py::TestEnpointsAvailability::test_index PASSED           [ 75%]
#10 0.772 tests/test_api.py::TestEnpointsAvailability::test_predict PASSED         [100%]
#10 0.776
#10 0.776 ============================== 4 passed in 0.24s ===============================
#10 DONE 0.8s

You are good if all tests are passing.

Model

Same as api, run:

$ cd model/
$ docker build -t model_test --progress=plain --target test .