Docker Compose deployment for the CoGS web service

Modifying the database

The following command will open a Python REPL in the context of the backend:

$ docker exec -ti ${env}_backend_1 python3

(where ${env} is either development or testing, depending on which environment you are attempting to interact with)

This can then be used as described in the backend README to interact with the database via the Python ORM.

Development

The basic idea: the master branch of this repository (and its submodules) should always be in a fit state to deploy to production.

Development happens (ideally in branches) in the frontend and backend repositories, with development instances being run from checkouts of whatever branches are relevant; at the point that a new feature lands in the master branch of one or both of those repositories (having been tested before merging to master), the update-submodules.sh script can be used to update this repository with the new commits from the submodules, and the production instance can then be updated from a checkout of this repository.

In other words, development looks like this:

~/deploy/backend (master)$ git checkout -b my-feature-branch
~/deploy/backend (my-feature-branch)$ # implement the feature
~/deploy/backend (my-feature-branch)$ git commit -m "Implement feature"
~/deploy/backend (my-feature-branch)$ # test the feature
~/deploy/backend (my-feature-branch)$ git checkout master
~/deploy/backend (master)$ git merge my-feature-branch
~/deploy/backend (master)$ git push
~/deploy/backend (master)$ cd ~/deploy
~/deploy (master)$ ./update-submodules.sh
Updated submodules and pushed to remote

then, on the production server, deployment looks like this:

~/deploy (master)$ # get the new commits
~/deploy (master)$ git pull --recurse-submodules=on-demand
~/deploy (master)$ # checkout the submodules to the correct commits
~/deploy (master)$ git submodule update --recursive