A template for docker based Yii 2 applications.
- Ephemeral container, configured via environment variables
- Application specific base image (Nginx + PHP-FPM)
- Optional local configuration overrides for development/debugging (git-ignored)
- Base scaffold code for login, signup and forgot-password actions
- Flat configuration file structure
- Optional cron integration for periodic jobs
Note: The included example base image is now based on Alpine Linux and uses s6-overlay to supervise Nginx + PHP-FPM. You can of course change this to any setup you prefer. You find the "old" setup based on Apache and mod_php in the "apache" branch. Note though, that it's no longer maintained.
The core idea of this template is that you build a bespoke base image for your application that meets your project's exact requirements. This image contains:
- PHP runtime environment (e.g. Nginx + PHP-FPM)
- PHP extensions
- Composer packages
The base image will hardly ever change unless
- you want to upgrade to a newer PHP or Yii version or
- you want to add a missing PHP extension or composer package.
Its configuration can be found in the ./build
directory:
Dockerfile
adds PHP extensions and required system packagescomposer.json
andcomposer.lock
list composer packages
The actual app image extends from this base image and uses ./Dockerfile
in the main directory. It basically only adds your app sources and productive
PHP config on top.
In the recommended scenario you would build the base image once then upload it to your container registry and share it with your co-developers.
If you don't have a container registry you can still use our template. But then each developer in your team will have to build the same base image locally.
We follow docker's principle that containers are configured via environment
variables. This only includes runtime configuration, of course: Things like
whether to run in debug mode or DB credentials. Most other settings like for
example URL rules will be hardcoded in ./config/web.php
.
You should continue to follow this principle when developing your app. For more details also see our yii2-configloader that we use in this template.
Note: There's also one important main setting for php-fpm that affects how many children should be started. This depends on the RAM you have available. See
PHP_FPM_MAX_CHILDREN
indocker-compose-example.yml
.
First fetch a copy of our application template, for example with git:
git clone https://github.com/codemix/yii2-dockerized.git myproject
rm -rf myproject/.git
You could also download the files as ZIP archive from GitHub.
To manage composer packages We use a container that is based on the official
composer image. First go to the
./build
directory of the app:
cd myproject/build
To add a package run:
docker-compose run --rm composer require some/library
To update all packages run:
docker-compose run --rm composer update
This will update composer.json
and composer.lock
respectively. You can
also run other composer commands, of course.
You now have to rebuild your base image! (see below).
Note: As docker's composer image may not meet the PHP requirements of all your packages you may have to add
--ignore-platform-reqs
to be able to install some packages.
Before you continue with building the base image you should:
- Set a tag name for the base image in
./build/docker-compose.yml
- Use the same tag name in
./Dockerfile
in the main directory - Optionally add more PHP extensions or system packages in
./build/Dockerfile
- Choose a timezone in
./build/Dockerfile
. This is only really relevant if you want to enable crond, to let the jobs run at correct local times.
To start you first need to create an initial composer.lock
. So go to the
./build
directory and run:
docker-compose run --rm composer install
Then you can build the base image:
docker-compose build
Now you could upload that image to your container registry.
At this point you may want to modify our application template, add some default configuration and remove those parts that you don't need. Afterwards you are ready for the initial commit to your project repository.
During development we map the local app directory into the app image. This way we always run the code that we currently work on.
As your local docker setup may differ from production (e.g. use different
docker network settings) we usually keep docker-compose.yml
out of version
control. Instead we provide a docker-compose-example.yml
with a reasonable
example setup.
Since the runtime configuration should happen with environment variables, we
use a .env
file. We also keep this out of version control and only include a
.env-example
to get developers started.
After you've cloned your project you need to prepare two files:
cd myproject
cp docker-compose-example.yml docker-compose.yml
cp .env-example .env
You should modify these two files e.g. to enable debug mode or configure a database. Then you should be ready to start your container and initialize your database (if your project has one):
docker-compose up -d
# Wait some seconds to let the DB container fire up ...
docker-compose exec web ./yii migrate
Finally you need to set write permissions for some directories. It's sufficient
if the www-data
group in the container has write permissions. This way your
local user can still own these directories:
docker-compose exec web chgrp www-data web/assets runtime var/sessions
docker-compose exec web chmod g+rwx web/assets runtime var/sessions
When done, you can access the new app from http://localhost:8080.
A development session will usually go like this:
docker-compose up -d
# edit files, check, tests, ...
git add .
git commit -m 'Implemented stuff'
docker-compose stop
Note: Another approach is to leave a terminal window open and start the container with
docker-compose up
This way you can always follow the live logs of your container. To stop all containers, press
Ctrl-c
.
If you need to run yiic commands you execute them in the running development container:
docker-compose exec web ./yiic migrate/create add_column_name
Note: If a command creates new files on your host (e.g. a new migration) you may have to change file permissions to make them writeable on your host system:
chown -R mike migrations/*
The procedure for adding or updating composer packages is the same as described in 2.2 above. Remember that you have to rebuild the base image afterwards. It should probably also receive an updated version tag.
Sometimes you may have to add complex parts to config/web.php
but want
to avoid the risk of accidentally committing them. You therefore can activate
support for a config/local.php
file in your .env
. This config file will
be merged into config/web.php
.
# Whether to load config/(local|console-local).php. Default is 0.
ENABLE_LOCALCONF=1
For some IDEs it's useful to have the composer packages available on your local host system. To do so you can simply copy them from inside the container:
docker-compose exec web cp -rf /var/www/vendor ./
Note: Inside the container composer packages live in
/var/www/vendor
instead of the app's./vendor
directory. This way we don't override the vendor directory when we map the local app directory into the container.
To run periodic jobs the integrated crond (busybox implementation) can be
activated by setting ENABLE_CROND=1
in the docker-compose.yml
file.
Cron jobs are added in config/crontabs/<username>
. There's an example file
for www-data
included. When changing a file there the container must be
restarted to activate the crontab.