SML Portal

This repo contains the build environment and code to generate and upload the Statistical Methods Library (SML) Portal Web Application.

Prerequisites

Before continuing, install the following tools:

Homebrew: Follow the instructions on this page

Python: To install the correct version of python on your local system, check the pyproject.toml and then install the relevant version.

You will need to install poetry to install the python dependencies. If you want to manage multiple versions of python you will need to install pyenv.

Refer to the confluence page for guidance on installing and using pyenv and poetry.

You will also need to have installed wget to fetch the ONS Design System.

If you have not set up a GPG key with your GitHub account you will need to also follow these instructions on installing a gpg key before you can make a commit.

Build and Deploy

Fetch the ONS Design System

Download the release of the ONS Design System, and unpack them into the correct location with this command:

./get_design_system.sh

Make the site

You will need to install the python dependencies, including Frozen-Flask, the static website generator:

poetry install --sync

You can activate your virtual environment with the following command:

poetry shell

With Jsonnet content in the content/ directory, you should now be able to run the Flask demo server:

poetry run flask --app sml_builder --debug run

If everything runs without errors, you should now be able to navigate to http://127.0.0.1:5000/ to view the site.

Linters & Security Scan

To run the linters (Pylint & Flake8) and security scan (Bandit):

sudo ./run_py_tools.sh

Troubleshooting

Access denied message when trying to reach CloudFront URL

• One common cause of this error message is not having uploaded your site yet.
• Are you coming from a non-UK IP address, possibly due to a VPN being active? If so, switch this off and try again. Geographical restrictions are in place.

If you have any errors running the web app or any other errors, it maybe because of issues relating to your latest macOS version and Xcode command line tools installed. You can try the command below where you will remove and reinstall Xcode command line tools:

sudo rm -rf /Library/Developer/CommandLineTools
xcode-select --install
/usr/bin/xcodebuild -version

If you experience any problems launching the application check the version of macOS installed, as problems can arise if you are using macOS Monterey and following versions uses port 5000 for its control center causing a conflict.

If another program is already using port 5000, you may see the error OSError: [Errno 98] or OSError: [WinError 10013] when the server tries to start.

In this case run the following command, replacing it with the port you want the web app to run on, so that the application is launched on a different port.

poetry run flask --app sml_builder --debug run --port=8000

Behaviour tests

The selenium behaviour tests can be run using the following command. If you are running the web app on a different port than port 5000, please change the port number below before running the command.

behave -D host=http://127.0.0.1:5000/

This will run the behaviour test locally in a headless state. If you want to see the GUI browser tests running then go to the setupSelenium.py file and comment the headless boolean.

Note: Investigation was undertaken to automate the behavior tests as part of the pipeline github action workflows.

However, we discovered the github agents are hosted in the United States and when visiting our services url, the runner routed to our 'page not found' webpage.

Hence as the page can only be accessed from within the uk we cannot automate the tests in the pipeline.

Concourse Setup

Prod and preprod are served as one pipeline which runs main

fly set-pipeline \
-t aws-sml \
-p live-sml-catalogue \
-c ci/live-pipeline.yml

While Dev is set as another

fly set-pipeline \
-t aws-sml \
-p dev-sml-catalogue \
-c ci/dev-pipeline.yml

To delete these run the commands

fly destroy-pipeline -t aws-sml -p live-sml-catalogue

and

fly destroy-pipeline -t aws-sml -p dev-sml-catalogue

When running the production pipeline we need to run it via fly set-pipeline in order to pass the tag version to deploy to production, this can be done as so:

fly set-pipeline \
-t aws-sml \
-p live-sml-catalogue \
-v RELEASE_CANDIDATE={VERSION} \
-c ci/live-pipeline.yml

Rollback of preprod and prod can be achieved via the rollback pipeline, run set-pipeline to pass the desired rollback tag version and then run the relevant environment via the concourse UI

fly set-pipeline \
-t aws-sml \
-p rollback-environment \
-v ROLLBACK_TAG={VERSION} \
-c ci/rollback-environment.yml

Note: Due to changes in the ci/tasks files to allow this functionality rollback is only available on versions beginning with v1.0.0 or later