django-hines

Code for https://www.gyford.com

Pushing to main will run the commit through this GitHub Action to run tests, and Coveralls to check coverage. If it passes, it will be deployed automatically to the VPS.

When changing the python version, it will need to be changed in:

.github/workflows/test.yml
.pre-commit-config.yaml
.python-version (for pyenv)
pyproject.toml (ruff's target-version)
docker/web/Dockerfile

For local development we use Docker. The live site is on an Ubuntu 22 VPS.

Local development setup

1. Create a .env file

Copy .env.dist to .env and alter any necessary settings.

2. Set up a local domain name

Open your /etc/hosts file in a terminal window by doing:

$ sudo vim /etc/hosts

Enter your computer's password. Then add this line somewhere in the file and save:

127.0.0.1 www.gyford.test

3. Build the Docker containers

Download, install and run Docker Desktop.

In same directory as this README, build the containers:

$ docker-compose build

Then start up the web, assets and database containers:

$ docker-compose up

There are five containers:

hines_web: the webserver
hines_db: the postgres server
hines_assets: the front-end assets builder
hines_redis: the redis server (for django-q2 and optional caching)
hines_djangoq: the django-q2 server

4. Set up the database

Once that's running, showing the logs, open another terminal window/tab.

There are two ways we can populate the database. First we'll create an empty one, and second we'll populate it with a dump of data from the live site.

4a. An empty database

The build step will create the database and run the initial Django migrations.

Then create a superuser:

$ ./run manage createsuperuser

(See below for more info on the ./run script.)

4b. Use a dump from the live site

Log into postgres and drop the current (empty) database:

$ ./run psql -d postgres
# drop database hines with (FORCE);
# create database hines;
# grant all privileges on database hines to hines;
# \q

On the VPS, create a backup file of the live site's database:

$ pg_dump dbname -U username -h localhost | gzip > ~/hines_dump.gz

Then scp it to your local machine:

$ scp [email protected]:/home/username/hines_dump.gz .

Put the file in the same directory as this README.

Import the data into the database ():

$ gunzip hines_dump.gz
$ docker exec -i hines_db pg_restore --verbose --clean --no-acl --no-owner -U hines -d hines < hines_dump

5. Vist and set up the site

Then go to http://www.gyford.test:8000 and you should see the site.

Log in to the Django Admin, go to the "Sites" section and change the one Site's Domain Name to www.gyford.test:8000 and the Display Name to "Phil Gyford’s website", if it's not already.

Ongoing work

Docker

Whenever you come back to start work you need to start the containers up again by doing this from the project directory:

$ docker-compose up

When you want to stop the server, then this from the same directory:

$ docker-compose down

You can check if anything's running by doing this, which will list any Docker processes:

$ docker ps

See details on the ./run script below for running things inside the containers.

Python dependencies with virtualenv and pip-tools

Adding and removing python depenencies is most easily done with a virtual environment on your host machine. This also means you can use that environment easily in VS Code.

Set up and activate a virtual environment on your host machine using virtualenv:

$ virtualenv --prompt . venv
$ source venv/bin/activate

We use pip-tools to generate requirements.txt from requirements.in, and install the dependencies. Install the current dependencies into the activated virtual environment:

(venv) $ python -m pip install -r requirements.txt

To add a new depenency, add it to requirements.in and then regenerate requirements.txt:

(venv) $ pip-compile --upgrade --quiet --generate-hashes

And do the pip install step again to install.

To remove a dependency, delete it from requirements.in, run that same pip-compile command, and then:

(venv) $ python -m pip uninstall <module-name>

To update the python dependencies in the Docker container, this should work:

$ ./run pipsync

But you might have to do docker-compose build instead?

pre-commit

Install pre-commit to run .pre-commit-config.yml automatically when git commit is done.

Front-end assets

Gulp is used to build the final CSS and JS file(s), and watches for changes in the hines_assets container. Node packages are installed and upgraded using yarn (see ./run below).

The ./run script

The ./run script makes it easier to run things that are within the Docker containers. This will list the commands available, which are outlined below:

$ ./run

`./run cmd`

Run any command in the web container. e.g.

$ ./run cmd ls -al

`./run sh`

Starts a Shell session in the web container.

`./run manage`

Run the Django manage.py file with any of the usual commands, within the pipenv virtual environment. e.g.

$ ./run manage makemigrations

The development environment has django-extensions installed so you can use its shell_plus and other commands. e.g.:

$ ./run manage shell_plus
$ ./run manage show_urls

`./run tests`

Runs all the Django tests. If it complains you might need to do ./run manage collecstatic first.

Run a folder, file, or class of tests, or a single test, something like this:

$ ./run tests tests.core
$ ./run tests tests.core.test_views
$ ./run tests tests.core.test_views.HomeViewTestCase
$ ./run tests tests.core.test_views.HomeViewTestCase.test_response_200

`./run coverage`

Run all the tests with coverage. The HTML report files will be at htmlcov/index.html.

`./run psql`

Conects to PosgreSQL with psql. Add any required arguments on the end. Uses the hines database unless you specify another like:

$ ./run psql -d databasename

`./run pipsync`

Update the installed python depenencies depending on the contents of requirements.txt.

`./run yarn:outdated`

List any installed Node packages (used for building front end assets) that are outdated.

`./run yarn:upgrade`

Update any installed Node packages that are outdated.

VPS set-up

The complete set-up of an Ubuntu VPS is beyond the scope of this README. Requirements:

Local postgresql
Local redis (for caching and django-q2)
pipx, virtualenv and pyenv
gunicorn
nginx
systemd
cron

1. Create a database

username$ sudo su - postgres
postgres$ createuser --interactive -P
postgres$ createdb --owner hines hines
postgres$ exit

2. Create a directory for the code

username$ sudo mkdir -p /webapps/hines/
username$ sudo chown username:username /webapps/hines/
username$ mkdir /webapps/hines/logs/
username$ cd /webapps/hines/
username$ git clone [email protected]:philgyford/django-hines.git code

3. ## Install python version, set up virtualenv, install python dependencies

username$ pyenv install --list  # All those available to install
username$ pyenv versions        # All those already installed and available
username$ pyenv install 3.10.8  # Whatever version we're using

Make the virtual environment and install pip-tools:

username$ cd /webapps/hines/code
username$ virtualenv --prompt hines venv -p $(pyenv which python)
username$ source venv/bin/activate
(hines) username$ python -m pip install pip-tools

Install dependencies from requirements.txt:

(hines) username$ pip-sync

4. Create `.env` file

(hines) username$ cp .env.dist .env

Then fill it out as required.

5. Set up database

Either do ./manage.py migrate and ./manage.py createsuperuser to create a new database, or import an existing database dumbp.

6. Set up gunicorn with systemd

Symlink the files in this repo to correct location for systemd:

username$ sudo ln -s /webapps/hines/code/conf/systemd_gunicorn.socket /etc/systemd/system/gunicorn_hines.socket
username$ sudo ln -s /webapps/hines/code/conf/systemd_gunicorn.service /etc/systemd/system/gunicorn_hines.service

Start the socket:

username$ sudo systemctl start gunicorn_hines.socket
username$ sudo systemctl enable gunicorn_hines.socket

Check the socket status:

username$ sudo systemctl status gunicorn_hines.socket

Start the service:

username$ sudo systemctl start gunicorn_hines

5. Set up nginx

Symlink the file in this repo to correct location:

username$ sudo ln -s /webapps/hines/code/conf/nginx.conf /etc/nginx/sites-available/hines

Enable this site:

username$ sudo ln -s /etc/nginx/sites-available/hines /etc/nginx/sites-enabled/hines

Remove the default site if it's not already:

username$ sudo rm /etc/nginx/sites-enabled/default

Check configuration before (re)starting nginx:

username$ sudo nginx -t

Start nginx:

 username$ sudo service nginx start

6. Set up django-q2 with systemd

Symlink the file in this repo to the correct location for systemd:

username$ sudo ln -s /webapps/hines/code/conf/systemd_djangoq.service /etc/systemd/system/djangoq_hines.service

Start the service:

username$ sudo systemctl start djangoq_hines

The tasks we have set up to use with django-q2

NOTE: If a task times out, it won't appear in the lists of Successful or Failed tasks.

Every 10 mins: hines.core.tasks.publish_scheduled_posts
Hourly: hines.core.tasks.fetch_flickr_photos, kwargs days="7", account="35034346050@N01"
Hourly: hines.core.tasks.fetch_lastfm_scrobbles, kwargs days="1", account="gyford"
Hourly: hines.core.tasks.fetch_pinboard_bookmarks, kwargs recent="20", account="philgyford"
Hourly: hines.core.tasks.fetch_twitter_tweets, kwargs recent="200", account="philgyford"
Hourly: hines.core.tasks.pending_mentions
Daily: hines.core.tasks.fetch_lastfm_scrobbles, kwargs days="14", account="gyford"
Daily: hines.core.tasks.fetch_twitter_favorites, kwargs recent="200", account="philgyford"
Daily: hines.core.tasks.fetch_twitter_files
Daily: hines.core.tasks.update_twitter_tweets, kwargs account="philgyford"
Daily: hines.core.tasks.update_twitter_users, kwargs account="philgyford"

Currently times out

Daily: hines.core.tasks.fetch_flickr_photosets, kwargs account="35034346050@N01" (took 1m 31s on command line)

Media files

Whether in local dev or Heroku, we need an S3 bucket to store Media files in (Static files are served using Whitenoise).

Go to the IAM service, Users, and 'Add User'.
Enter a name and check 'Programmatic access'.
'Attach existing policies directly', and select 'AmazonS3FullAccess'.
Create user.
Save the Access key and Secret key.
On the list of Users, click the user you just made and note the User ARN.
Go to the S3 service and 'Create Bucket'. Name it, select the region, and click through to create the bucket.
Click the bucket just created and then the 'Permissions' tab. Add this policy, replacing BUCKET-NAME and USER-ARN with yours:

{
  "Statement": [
    {
      "Sid": "PublicReadForGetBucketObjects",
      "Effect": "Allow",
      "Principal": {
        "AWS": "*"
      },
      "Action": ["s3:GetObject"],
      "Resource": ["arn:aws:s3:::BUCKET-NAME/*"]
    },
    {
      "Action": "s3:*",
      "Effect": "Allow",
      "Resource": ["arn:aws:s3:::BUCKET-NAME", "arn:aws:s3:::BUCKET-NAME/*"],
      "Principal": {
        "AWS": ["USER-ARN"]
      }
    }
  ]
}

Click on 'CORS configuration' and add this:

<CORSConfiguration>
<CORSRule>
    <AllowedOrigin>*</AllowedOrigin>
    <AllowedMethod>GET</AllowedMethod>
    <MaxAgeSeconds>3000</MaxAgeSeconds>
    <AllowedHeader>Authorization</AllowedHeader>
</CORSRule>
</CORSConfiguration>

Upload all the files to the bucket in the required location.
Update the server's environment variables for AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_STORAGE_BUCKET_NAME.

Image cache

To clear the cached thumbnail images created by django-imagekit (used by django-spectator):

Delete all the images from the CACHES directories on S3.
Clear the Redis cache, as above.

To re-generate all the cached thumbnail images (which must be done because of the "Optimistic" cache file strategy):

(hines) username$ ./manage.py generateimages

django-hines django-hines copied to clipboard

Metadata