Skip to content

Latest commit

 

History

History
66 lines (42 loc) · 4.55 KB

aws_setup.md

File metadata and controls

66 lines (42 loc) · 4.55 KB
Raw

Instructions for using AWS S3 Static and Media Files

Guide: S3 Static & Media Files for Django
If you are using this project, then you won't have to do most of the steps as they have already been implemented.

AWS Setup

  1. Follow step 1 and step 2 from the guide

    • While creating an IAM User, there will be an option to add the user to a group. Click on that and create a new group.
    • Go to the Permissions tab and click on Attach Policy.
    • Paste the policy given in JSON format in the guide in step 5 and save it (Replace <your_bucket_name> in the policy with the name of the bucket that you'll create in step 3).
    • Then attach the policy to the group. These policies give the users in the group permissions to access the services.

  2. Follow step 3 from the guide

    • Select Region such that it has both versions 2 & 4 supported. In my case, the region is ap-southeast-1
      For reference, see: AWS Regions and Endpoints
    • Keep other settings as the default ones and create the bucket.

  3. Now go to the bucket created above and click on the Permissions tab. Go to CORS Configuration and click on Save (We will use the default configuration already written there).

Django Setup

  1. Follow step 1 till step 3 from the guide

    • boto and boto3 are python bindings for AWS. django-storages are used by django to send static files to AWS.
    • Add storages app to every settings module.
    • We migrate because we added a new app storages.

  2. Add the required credentials of the IAM user and AWS to kart/credentials.py (See kart/credentials-sample.py for reference).

  3. Follow the guide from step 4 till step 7

    • In step 4, create directory inside kart/ folder and add a init.py file to make it a module, and then create utils.py
    • Create the static and media folder in the bucket (created above) as specified in utils.py. Make those folders public in AWS. Create another folder in the bucket called protected/ (do not make it public), it can be used to store private static files.
    • In kart/aws/conf.py, update the S3DIRECT_REGION, app name (app name == kart) in DEFAULT_FILE_STORAGE, STATICFILES_STORAGE and bucket name in AWS_STORAGE_BUCKET_NAME.
    • set attribute AWS_QUERYSTRING_AUTH = True in conf.py
    • In step 7, do the import in local settings: kart/settings/local.py file.

  4. Now, test it in local. Run python manage.py collectstatic. It will collect all the static files and now it should be visible in AWS bucket.

  5. Now if we run the server in local, and inspect an image, the image source will give the AWS link.

  6. On opening the image source link, our access key and other info will be visible in the url. To disable that, set AWS_QUERYSTRING_AUTH = False in conf.py

  7. If we want to hide the static files, we can make the static folder in the AWS bucket to not public.

  8. Now test it in production, add the aws.conf import in production settings: kart/settings/production.py. We can now remove the static and media root stuff from production settings, but we keep it just to be on the safe side.

  9. After pushing the changes, migrate on the production server:
    heroku run python manage.py migrate

Note

  • Any existing media files (before the AWS setup) cannot be uploaded to the bucket automatically, they have to be manually uploaded either from AWS website, Django admin or through aws-cli. For cli, see: AWS CLI Setup
    Once AWS has been setup, all media files will be directly uploaded to AWS bucket.
  • Everytime any static or public-media file is added, the folder has to be made public again.
  • For testing, to disable the AWS upload and use the local static files, comment out the AWS conf.py import from the settings of all environments.

File Uploads

  1. To link the view with the uploaded files in AWS, one way is to link the view with the file by using the path of the file in AWS. This method is very useful in case of large uploads because if the file changes then only the file path has to be updated in the database.
    The other way is to use a Django FileField, this method is not suitable for large file uploads.

  2. If the same media file is uploaded to the same location, then by default AWS does not change its name and store it (like django does). AWS overwrites the previous file.