## Description
Following up on @webyneter attempt in #1205, which is now getting outdated, I've tried to make Gulp task runner work with Docker. There is no documentation yet, but this seems to work locally with the custom bootstrap compilation...
- [x] Add a node image for local developement
- [x] Proxy the django image rather than localhost in Browser sync, pass header to keep hostname
- [x] Don't call the runserver command from Gulp, let docker-compose handle
- [x] Update package.json & gulpfile.js templates to reduce the number of unwanted empty lines
- [x] Use [multi-stage build](https://docs.docker.com/develop/develop-images/multistage-build/) in production to make sure the static assets are produced
- [x] Update documentation
- [x] Verify that the previous issue with static assets missing from production isn't there
## Rationale
Currently, the static build isn't working nicely with Docker, extra manual setup is required.
Fixes#1762
## Description
Replace Caddy with Traefik
## Rationale
There is some trouble with the Caddy license (https://github.com/pydanny/cookiecutter-django/pull/1282#issuecomment-329617536)
@drdaeman suggested using Traefik (https://github.com/pydanny/cookiecutter-django/pull/1282#issuecomment-353655273) which supports ACME and also plays very nice with Docker.
## Comments
I am currently using the proposed setup on a live site and it working great so far. If this PR is of interest to the maintainers, then I could commit more changes and take care of the documentation. Of course, any suggestions by the more experienced people around here, are welcome!
[//]: # (Thank you for helping us out: your efforts mean great deal to the project and the community as a whole!)
[//]: # (Before you proceed:)
[//]: # (1. Make sure to add yourself to `CONTRIBUTORS.rst` through this PR provided you're contributing here for the first time)
[//]: # (2. Don't forget to update the `docs/` presuming others would benefit from a concise description of whatever that you're proposing)
## Description
[//]: # (What's it you're proposing?)
Added a note around CELERY_TASK_ALWAYS_EAGER = True in docker config for local development. This causes tasks to be executed on the 'main' thread rather than by the workers. I understand why that might be desirable, but thought it worth calling out incase (like me) it makes people think something is broken.
## Rationale
[//]: # (Why does the project need that?)
Ease of use/troubleshooting
## Use case(s) / visualization(s)
[//]: # ("Better to see something once than to hear about it a thousand times.")
- Mention the need for Redis if Celery is selected
- Link to PostgreSQL & Redis download pages
- Detail better how to set the environment
- Improve internal links using Sphinx' :ref
- Remove unused link
* Add note about using keep_local_envs_in_vcs
As a newbie, I wasn't sure about `keep_local_envs_in_vcs`, so I said yes, and when CC was building, it gave me the message:
"[INFO]: .env(s) are only utilized when Docker Compose and/or Heroku support is enabled so keeping them does not make sense given your current setup."
Seems like it could go in this documentation, and make things easier for newbies.
* Add me to CONTRIBUTORS.RST
* Integrate Flower with Docker Compose setup locally
* Remove alien worker celeryd option
* Move Flower COPY section below the worker's
* Remove set -o pipefail command from Flower start script
* Flower client authentication
* Override flower service image name
* Move flower service to the end of local.yml
* Install flower==0.9.2 in all environments
* Introduce production flower service
* Fix local flower start script
* Document Flower integration
* Prettify *.django envs
Rationale: consistency.
* Reference local environment Flower docs from the production's
* 'two more services' -> 'three more services'
* Export PG* envs when backing up postgres
* Export PG* envs when restoring postgres from backup
* Prevent postgres connection from dropping all at ones
* Alter postgres backups docs
Include another crucial prerequisite.
* "feel free switching" -> "feel free to switch"
* Address the feedback
* Update generated project's .gitignore
* Post-gen gitignore .env/ and .env
* Fix linesep between gitignored entries
* Persist `.env/**/*` files into cookiecutter-django's VCS
* Rename .env/ to .envs/
* Reference the newly created .envs/**/.* files in local.yml
* Reference the newly created .envs/**/.* files in production.yml
* Delete .env.example
* Refactor post-gen-project.py
Closes#1299.
* Implement production-dotenv-files-to-dotenv-file merge script
* Create shared PyCharm Run Configuration for the automation script
* Randomize POSTGRES_PASSWORD in ./envs/(.local|.production)/.postgres
* Default POSTGRES_PASSWORD and POSTGRES_USER to random values
* Fix jinja linebreaks in local.yml
* Spaces in production.yml
* Fix post-merge leftovers & set DJANGO_ADMIN_URL automatically
* Prettify here and there
* Fix FileNotFoundError
* Leave a TODO in post_gen_hook.py
* Introduce keep_local_envs_in_vcs option
* Remove envs when not opted for
* Inline pre_gen_project.py if-condition
* Get rid of PROJECT_DIR_PATH in post_gen_project.py
* Clean up the docs
* Match copyright notices
* Document envs ins and outs
When creating a Mailgun add-on on Heroku, the app gets some environment
variables by default:
MAILGUN_API_KEY
MAILGUN_DOMAIN
However, the cookiecutter names do not match and requires a manual step
from the user deploying. It's used elsewhere but shouldn't harm the
other deployment methods to rename these variables.
While updating the docs I noticed a variable that appear unused
DJANGO_MAILGUN_SERVER_NAME so this removes it from the documentation.
* Fix ./hooks/pre_gen_project.py asking user to select an option once only
+ prettify output
* Fix pre_gen hook not really exiting when it should
* Refactor & prettify ./hooks/post_gen_project.py
* Ensure same POSTGRES_USER is set across environments
+ get rid of env.example in favor of pre-generated .env.
* Add Bootstrap to package.json in case of custom_bootstrap_compilation
* Update JS task runners and base HTML to handle custom scss compilation
* Generate a vendors.js with custom bootstrap compilation + Gulp
* Update documentation accordingly
* Add missing if/endif in gulpfile
* Update to Bootstrap v4 final
* Introduce static asset build infrastructure
* Enhance gulpfile.js
* Introduce node service
* BrowserSync debug-only support
* Remove newline before BrowserSync debug-only support section
* FIx node Dockerfile package.json COPY
* Try fiixing node Dockerfile package.json COPY ones again
* Switch to `node:7-slim`
* Try switching to node:6
To account for possible node:7 docker-compose incompatibiltiy
* Revert "Try switching to node:6"
This reverts commit 62cc02df1a.
* Try switcging workdir to /app
* Try utilizing relative package.json path
* Resetting to the last version working locally with docker-compose 1.11.x
* Build upon the latest node:7.9-slim
* Stop dockerignoring package.json
* Fix typo
* Try a different package.json path
* Revert "Try a different package.json path"
This reverts commit f29f8500b8.
* Revert "Fix typo"
This reverts commit 02033729b5.
* Revert "Stop dockerignoring package.json"
This reverts commit 63c5491546.
* Upgrade docker-engine and docker-compose used by Travis CI
* Fix .travis.yml comments
* Inline docker-engine and docker-compose versions
* DEBUG: pwd
* Revert "DEBUG: pwd "
This reverts commit 6c2ed4321a.
* Try copying package.json to the same dir as node Dockerfile's
* Revert "Try copying package.json to the same dir as node Dockerfile's"
This reverts commit 24340a0783.
* Try out node:7.9
* Revert "Try out node:7.9"
This reverts commit 32286d33c2.
* Revert "Upgrade docker-engine and docker-compose used by Travis CI"
* Get rid of npm-check-updates
Reason: Reserved for the upcoming PR
* Get rid of npm-check
Reason: Reserved for the upcoming PR
* Get rid of 'standard' npm package
Reason: Reserved for the upcoming PR
* Clean up package.json
* Preserve package.json uncoditionally
Since we now have *unconditional* node.js integration, `package.json` must be out there whenever `node` service gets built
* Upgrade node service image to 7.10
* Document Node.js-Docker integration
* Fix gulpfile.js images region name
* Get rid of Gulp migrate task
* Document Gulp-Docker integration
* Introduce static asset build infrastructure
* Enhance gulpfile.js
* Introduce node service
* BrowserSync debug-only support
* Remove newline before BrowserSync debug-only support section
* FIx node Dockerfile package.json COPY
* Try fiixing node Dockerfile package.json COPY ones again
* Switch to `node:7-slim`
* Try switching to node:6
To account for possible node:7 docker-compose incompatibiltiy
* Revert "Try switching to node:6"
This reverts commit 62cc02df1a.
* Try switcging workdir to /app
* Try utilizing relative package.json path
* Resetting to the last version working locally with docker-compose 1.11.x
* Build upon the latest node:7.9-slim
* Stop dockerignoring package.json
* Revert "Stop dockerignoring package.json"
This reverts commit 63c5491546.
* Fix typo
* Revert "Fix typo"
This reverts commit 02033729b5.
* Try a different package.json path
* Revert "Try a different package.json path"
This reverts commit f29f8500b8.
* Upgrade docker-engine and docker-compose used by Travis CI
* Fix .travis.yml comments
* Inline docker-engine and docker-compose versions
* DEBUG: pwd
* Revert "DEBUG: pwd "
This reverts commit 6c2ed4321a.
* Try copying package.json to the same dir as node Dockerfile's
* Revert "Try copying package.json to the same dir as node Dockerfile's"
This reverts commit 24340a0783.
* Try out node:7.9
* Revert "Try out node:7.9"
This reverts commit 32286d33c2.
* Revert "Upgrade docker-engine and docker-compose used by Travis CI"
* Get rid of npm-check-updates
Reason: Reserved for the upcoming PR
* Get rid of npm-check
Reason: Reserved for the upcoming PR
* Get rid of 'standard' npm package
Reason: Reserved for the upcoming PR
* Clean up package.json
* Preserve package.json uncoditionally
Since we now have *unconditional* node.js integration, `package.json` must be out there whenever `node` service gets built
* Upgrade node service image to 7.10
* Document Node.js-Docker integration
* Fix gulpfile.js images region name
* Get rid of Gulp migrate task
* Document Gulp-Docker integration
* Remove Gulp-Docker integraton not supported initialization message
ADDED: HTTPS is on by default. This will give a new user an
understanding of why Cookie Django is set up securely for deployment in
a production environment.
Commit 04a58d5 removed the Heroku instant deploy button but the docs
still mention this as an option, this commit updates the docs to reflect
this change.
Includes:
* First pass at Elastic Beanstalk integration
* Gets code and elasticache working
* Very rudimentary documentation
* Includes post hook cleanup
Removed sections about using Docker Toolbox with virtualbox and added links to Docker for Mac/Windows.
Docker for Mac/Windows uses native virtual machines instead of virtualbox and introduces many improvements.
See: https://blog.docker.com/2016/03/docker-for-mac-windows-beta/Resolves: #706
* Remove webpack and merge
* Put postgresql_version line back in cookiecutter.json
* Put goldhand back in contributors file, added ssteinerX
* Add *.egginfo to .gitignore
* Fix dangling endif in README.rst
In order to resolve the name through docker's dns, we have to specify
the usual container name (certbot) - docker-compose only does
automatically this if you use `start certbot`, not with `run`.
Also added --rm to remove the container after it's done so we don't have
multiple with the same name
* Add webpack as an option
Adds webpack as a js_taskrunner option to cookiecutter-json. Will clone @hzdg/cookiecutter-webpack --pydanny-django branch into the project using cookiecutter's api in post_hooks.
The static webpack project will be placed into the <project_slug>/static/<project_slug>/ directory.
The webpack configs are placed in the ./config/ directory.
The cookiecutter-webpack project includes react / redux / karma configurations that are brought into the project.
* Add webpack documentation
As somebody new to mailhog, I was a confused on how to use it. I first assumed, it was going to be used by the django app itself. After reading more on the internet I found it needed to be executed outside of the django application.
The app name wasn't displayed correctly on the new relic dashboard.
Although `NEW_RELIC_APP_NAME` was set in `production.py`, the variable was not loaded *before* the newrelic agent was called.
Since the new relic agent is loaded at a very low level, NEW_RELIC_APP_NAME has to be set before django settings are imported.
