Manala Manala

Recipes for your projects

Elao - App - Docker

Migrations

Requirements

MacOS

  • Docker Desktop 4.3.2+ (brew install docker)
  • Mutagen Compose 0.13.0+ (brew install mutagen-io/mutagen/mutagen-compose)

Linux

Overview

This recipe contains some helpful scripts in the context of a php/nodejs app, such as Makefile tasks in order to release and deploy your app.

Init

cd [workspace]
manala init -i elao.app.docker [project]

Quick start

In a shell terminal, change directory to your app, and run the following commands:

cd /path/to/my/app
manala init
Select the "elao.app.docker" recipe

Edit the Makefile at the root directory of your project and add the following lines at the beginning of the file:

.SILENT:

-include .manala/Makefile

Then update the .manala.yaml file (see the deliveries example below) and then run the manala up command:

manala up

Warning

Don't forget to run the manala up command each time you update the .manala.yaml file to actually apply your changes !!!

From now on, if you execute the make help command in your console, you should obtain the following output:

Usage: make [target]

Help:
  help This help

Docker:
  docker Run docker container

App:

Environment interaction

In your app directory.

Initialise your app:

make setup

Start environment:

make up

Stop environment:

make halt

Environment shell:

make sh

Custom docker compose command:

make docker [COMMAND]

⚠︎ separate hyphen based arguments or flags with -- to avoid shell miss-interpretation:

make docker logs -- --follow

Configuration

Here is an example of a configuration in .manala.yaml:

###########
# Project #
###########

project:
    name: app
    ports_prefix: 123 # >= 20, <= 640

##########
# System #
##########

system:
    version: 11
    #timezone: Etc/UTC # Optional
    #locales: # Optional
    #    default: C.UTF-8
    #    codes: []
    #env: # Optional
    #    FOO: bar
    apt:
        #repositories: [] # Optional
        #preferences: [] # Optional
        packages:
          - pdftk
          - https://github.com/wkhtmltopdf/packaging/releases/download/0.12.6-1/wkhtmltox_0.12.6-1.buster_{{ apt_architecture }}.deb
    files:
      - path: /srv/app/var/log
        src: /srv/log
        state: link_directory
      - path: /srv/app/var/cache
        src: /srv/cache
        state: link_directory
      #- path: /srv/app/var/sessions
      #  src: /srv/sessions
      #  state: link_directory
      ## Api
      #- path: /srv/app/api/var/log
      #  src: /srv/log/api
      #  state: link_directory
      #- path: /srv/app/api/var/cache
      #  src: /srv/cache/api
      #  state: link_directory
    #network: # Optional
    #    hosts:
    #        127.0.0.1: foo.fr foobar.fr
    nginx:
        configs: []
        # configs:
        #     - template: nginx/gzip.j2
        #     #- template: nginx/cors.j2
        #     #- template: nginx/no_index.j2
        #     - template: nginx/php_fpm_app.j2
        #     # App
        #     - file: app.conf
        #       config: |
        #           server {
        #               server_name ~.;
        #               root /srv/app/public;
        #               access_log /srv/log/nginx.access.log;
        #               error_log /srv/log/nginx.error.log;
        #               include conf.d/gzip;
        #               location / {
        #                   try_files $uri /index.php$is_args$args;
        #               }
        #               location ~ ^/index\.php(/|$) {
        #                   include conf.d/php_fpm_app;
        #                   internal;
        #               }
        #           }
    php:
        version: 8.1
        # composer:
        #   version: 1 # Optional
        extensions:
          # Symfony
          - intl
          - curl
          - mbstring
          - xml
          # App
          - mysql
        configs:
          - template: php/opcache.ini.j2
          - template: php/app.ini.j2
            config:
                date.timezone: UTC
                upload_max_filesize: 16M
                post_max_size: 16M
    nodejs:
        version: 16
        # packages:
        #   - package: mjml
        #     version: 4.6.3
        # yarn:
        #     version: 1
    # cron:
    #     files:
    #       - file: app
    #         env:
    #           HOME: /srv/app
    #         jobs:
    #           # Foo - Bar
    #           - command: php bin/console app:foo:bar --no-interaction -vv >> /srv/log/cron.foo-bar.log 2>&1
    #             minute: 0
    #             # Dev
    #             state: absent
    # supervisor:
    #     configs:
    #       - file: app.conf
    #         #groups:
    #         #    acme:
    #         #        programs:
    #         #          - foo
    #         #          - bar
    #         programs:
    #             foo:
    #                 command: php bin/console app:acme:foo --no-interaction -vv
    #                 directory: /srv/app
    #                 stdout_logfile: /srv/log/supervisor.acme-foo.log
    #             bar:
    #                 command: php bin/console app:acme:bar --no-interaction -vv
    #                 directory: /srv/app
    #                 stdout_logfile: /srv/log/supervisor.acme-bar.log
    #             foo-bar:
    #                 command: php bin/console app:foo:bar --no-interaction -vv
    #                 directory: /srv/app
    #                 stdout_logfile: /srv/log/supervisor.foo-bar.log
    #       - file: app_foo.conf
    #         config: |
    #             [program:foo]
    #             command=/bin/foo
    # MariaDB
    mariadb:
        version: 10.6
    # ...*OR* MySQL...
    mysql:
        version: "8.0"
    # redis:
    #     version: "*"
    #     # config:
    #     #     save: '""' # Disable persistence
    elasticsearch:
        version: 7
        plugins:
          - analysis-icu
    # influxdb:
    #     version: "*"
    #     config:
    #       reporting-disabled: true
    #     databases:
    #       - app
    #     users:
    #       - database: app
    #         name: app
    #         password: app
    #     privileges:
    #       - database: app
    #         user: app
    #         grant: ALL
    # mongodb:
    #     version: 4.4
    ssh:
        client:
            config: |
                Host *.rix.link
                    User app
                    ForwardAgent yes
    docker:
        services:
            whoami:
                image: traefik/whoami:v1.7.1
                network_mode: service:app
                profiles:
                    - development
            # App ports
            app:
                ports:
                    # whoami
                    - 12345:80
        # Optimizes Mutagen sync performances (adapt to your project structure)
        mutagen:
            ignore:
                paths:
                    # Webpack build files
                    - /public/build/
                    # Symfony log & cache files
                    - /var/cache
                    - /var/log

Details:

  • project
    • ports_prefix: docker network behavior force localhost usage for all projects. In order to runs multiple projects simultaneously, a kind of range ports must be set to avoid conflicts. Choose a prefix value, greater or equal to 20, and lower or equal to 640, like 123. All project ports will be based on this value, like 12380 for http or 12343 for https accesses.

Integration

Jenkins

Here are some examples of integration configurations in .manala.yaml for Jenkins:

###############
# Integration #
###############

integration:
    jenkins:
        pipeline:
            tasks:
              - shell: make install@integration
              - label: Integration
                junit: report/junit/*.xml
                parallel: true
                warn: true
                tasks:
                  - label: Lint
                    tasks:
                      - shell: make lint.php-cs-fixer@integration
                      - shell: make lint.twig@integration
                      - shell: make lint.yaml@integration
                      - shell: make lint.eslint@integration
                  - label: Security
                    tasks:
                      - shell: make security.symfony@integration
                      - shell: make security.yarn@integration
                  - label: Test
                    tasks:
                      - shell: make test.phpunit@integration
                        artifacts: var/log/*.log
                        # artifacts:
                        #   - var/log/foo.log
                        #   - var/log/bar.log
                        # env:
                        #     DATABASE_URL: mysql://root@127.0.0.1:3306/app

In this example we have two parallel stages: api and mobile, corresponding to two different sub-apps.

###############
# Integration #
###############

integration:
    jenkins:
        pipeline:
            tasks:
              - label: Integration # Optional
                parallel: true # ! Careful ! Could *NOT* be nested !
                junit: report/junit/*.xml
                artifacts: var/log/*.log
                warn: true # Turn errors into warnings (recursively applied)
                tasks:
                  - app: api # Optional
                    tasks:
                      - shell: make install@integration
                      - shell: make build@integration
                      - shell: make lint.php-cs-fixer@integration
                      - shell: make security.symfony@integration
                      - shell: make test.phpunit@integration
                        artifacts: var/log/*.log
                        # env:
                        #     DATABASE_URL: mysql://root@127.0.0.1:3306/app
                  - app: mobile
                    tasks:
                      - shell: make install@integration
                      - shell: make build@integration
                      - shell: make lint.eslint@integration
                      - shell: make test.jest@integration

Github Actions

The recipes can generate Github actions files you can use in your workflows.
Complete the integration.github.jobs according to your need, and consult the .manala/github/integration/README.md to learn how you can write your own workflows using these actions.

###############
# Integration #
###############

integration:
  github:
    jobs:
      # Lint
      lint:
        tasks:
          - shell: make install@integration
          - shell: make build@integration
          - shell: make lint.twig@integration
          - shell: make lint.yaml@integration
          - shell: make lint.phpstan@integration
          - shell: make lint.php-cs-fixer@integration
          - shell: make lint.eslint@integration
      # Security
      security:
        tasks:
          - shell: make security.symfony@integration
          - shell: make security.npm@integration
      # Test
      test:
        tasks:
          - label: Install
            shell: |-
              make install.php@integration
              # Fake manifest.json
              mkdir -p public/build/
              echo {} > public/build/manifest.json
              # Install phpunit
              vendor/bin/simple-phpunit install
          - shell: make test.phpunit@integration
        artifacts:
          - var/log/test.log

Common integration tasks

Add in your Makefile:

###########
# Install #
###########

...

install@integration: export APP_ENV = test
install@integration:
    # Composer
    composer install --ansi --verbose --no-interaction --no-progress --prefer-dist --optimize-autoloader --no-scripts --ignore-platform-reqs
    #composer run-script symfony-scripts --ansi --verbose --no-interaction
    # Npm
    npm install --color=always --no-progress --no-audit
    # Yarn
    yarn install --color=always --no-progress

#########
# Build #
#########

...

build@integration:
    # Webpack Encore
    npx encore production --color=always --no-progress

########
# Lint #
########

...

lint.php-cs-fixer@integration:
    mkdir -p report/junit
    vendor/bin/php-cs-fixer fix --dry-run --diff --format=junit > report/junit/php-cs-fixer.xml

lint.phpstan@integration:
    mkdir -p report/junit
    vendor/bin/phpstan --error-format=junit --no-progress --no-interaction analyse > report/junit/phpstan.xml

lint.twig@integration:
    bin/console lint:twig templates --ansi --no-interaction

lint.yaml@integration:
    bin/console lint:yaml config translations --ansi --no-interaction

lint.eslint@integration:
    npx eslint src --format junit --output-file report/junit/eslint.xml

lint.stylelint@integration:
    mkdir -p report/junit
    npx stylelint "assets/styles/**/*.scss" \
        --syntax scss \
        --custom-formatter "node_modules/stylelint-junit-formatter" \
        > report/junit/stylelint.xml

lint.flow@integration:
    mkdir -p report/junit
    npx flow check --json | npx flow-junit-transformer > report/junit/flow.xml

############
# Security #
############

...

security.symfony@integration:
    symfony check:security

security.yarn@integration:
    yarn audit ; RC=$${?} ; [ $${RC} -gt 2 ] && exit $${RC} || exit 0

security.npm@integration:
    npm audit --audit-level moderate

########
# Test #
########

...

test.phpunit@integration: export APP_ENV = test
test.phpunit@integration:
    # Db
    bin/console doctrine:database:create --ansi
    bin/console doctrine:schema:create --ansi
    # PHPUnit
    bin/phpunit --colors=always --log-junit report/junit/phpunit.xml

test.jest@integration: export JEST_JUNIT_OUTPUT_DIR = report/junit
test.jest@integration: export JEST_JUNIT_OUTPUT_NAME = jest.xml
test.jest@integration:
    npx jest --ci --color --reporters=default --reporters=jest-junit

Deliveries

Here is an example of a production/staging deliveries configuration in .manala.yaml:

##############
# Deliveries #
##############

deliveries:

  - &delivery
    #app: api # Optional
    tier: production
    #ref: staging # Default to master
    # Release
    release_repo: git@git.example.com:<vendor>/<app>-release.git
    #release_ref: master # Based on app/tier by default
    # Whether to markup releases on original repository.
    # If true, a commit referencing the release repository commit will be pushed to the original repository.
    #release_markup: true
    release_tasks:
      - shell: make install@production
      - shell: make build@production
    # You can either explicitly list all the paths you want to include
    release_add:
      - bin
      - config
      - public
      - src
      - templates
      - translations
      - vendor
      - composer.* # Composer.json required by src/Kernel.php to determine project root dir
                   # Composer.lock required by composer on post-install (warmup)
      - Makefile

    # Or you can include all by default and only list the paths you want to exclude
    # release_remove:
    #   - ansible
    #   - build
    #   - doc
    #   - node_modules
    #   - tests
    #   - .env.test
    #   - .php_cs.dist
    #   - .manala*
    #   - package.json
    #   - phpunit.xml.dist
    #   - README.md
    #   - webpack.config.js
    #   - yarn.lock

    # Deploy
    deploy_hosts:
      - ssh_host: foo-01.bar.elao.local
        #master: true # Any custom variable are welcomed
      - ssh_host: foo-02.bar.elao.local
    deploy_dir: /srv/app
    #deploy_url: https://foo.com # Url to visit after deploy
    deploy_shared_files:
      - config/parameters.yml
    deploy_shared_dirs:
      - var/log
    deploy_tasks:
      - shell: make warmup@production
      #- shell: make migration@production
      #  when: master | default # Conditions on custom host variables (jinja2 format)
    #deploy_remove:
    #  - web/app_dev.php
    deploy_post_tasks:
      - shell: sudo /bin/systemctl reload php8.1-fpm
      #- shell: sudo /bin/systemctl restart supervisor
    # GitHub
    github_ssh_key_secret: SSH_DEPLOY_KEY_PRODUCTION

  - << : *delivery
    tier: staging
    release_tasks:
      - shell: make install@staging
      - shell: make build@staging
    # Deploy
    deploy_hosts:
      - ssh_host: foo.bar.elao.ninja.local
    deploy_tasks:
      - shell: make warmup@staging
    # GitHub
    github_ssh_key_secret: SSH_DEPLOY_KEY_STAGING

Github Actions

Deliveries can be triggered through Github Actions as well. Consult the .manala/github/deliveries/README.md to learn how to write your own release & deploy workflows.

Note

The github_ssh_key_secret key must be set on each delivery, with the Github Secret key that holds the SSH private key used to access the release repository and the deployment server.

Additionally, some trigger.* Make targets are generated to trigger Github workflows from your terminal.
It uses the Github CLI tool to make API calls, so be sure to execute:

gh login auth

at least once on your machine, if not already.

Makefile

Makefile targets that are supposed to be runned via docker must be prefixed.

foo: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
foo:
    # Do something really foo...

Ssh

#######
# Ssh #
#######

## Ssh to staging server
ssh@staging: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
ssh@staging:
    ssh app@foo.staging.rix.link

# Single host...

ssh@production: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
ssh@production:
    ...

# Multi host...

ssh@production-01: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
ssh@production-01:
    ...

Sync

sync@staging: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
sync@staging:
    mkdir -p var
    rsync --archive --compress --verbose --delete-after \
        app@foo.staging.rix.link:/srv/app/current/var/files/ \
        var/files/

# Multi targets...
sync-uploads@staging: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
sync-uploads@staging:
  ...

# Multi apps...
sync.api-uploads@staging: SHELL := $(or $(DOCKER_SHELL),$(SHELL))
sync.api-uploads@staging:
  ...

Git tools

This recipe contains some git helpers such as the git_diff function.

This function is useful for example to apply php-cs, php-cs-fix or PHPStan checks only on the subset of updated PHP files and not on any PHP file of your project.

Usage (in your Makefile):

lint.php-cs-fixer: DIFF = $(call git_diff, php, src tests)
lint.php-cs-fixer:
    $(if $(DIFF), \
        vendor/bin/php-cs-fixer fix --config=.php_cs.dist --path-mode=intersection --diff --dry-run $(DIFF), \
        printf "You have made no change in PHP files\n" \
    )

Try tools

This recipe contains some try helpers such as the try_finally function.

This function is useful for example to run phpunit tests needing a started symfony server, and to stop this server regardless of the tests return code.

Usage (in your Makefile):

test.phpunit@integration:
    symfony server:start --ansi --no-humanize --daemon --no-tls --port=8000
    $(call try_finally, \
        bin/phpunit --colors=always --log-junit report/junit/phpunit.xml, \
        symfony server:stop --ansi \
    )

Secrets

In order to generate secrets, use Gomplate, called by a make task. Gomplate takes a template, queries its values from a Vault server and renders a file.

Add the following tasks in the Makefile:

###########
# Secrets #
###########

secrets@production:
    gomplate --input-dir=secrets/production --output-map='{{ .in | replaceAll ".gohtml" "" }}'

secrets@staging:
    gomplate --input-dir=secrets/staging --output-map='{{ .in | replaceAll ".gohtml" "" }}'

Put your templates in .gohtml files inside a secrets/[production|staging] directory at the root of the project.
Respect destination file names, extensions, and paths:

secrets
├── production
|   ├── .env.gohtml
|   └── config
|       └── parameters.yaml.gohtml
└── staging
    ├── .env.gohtml
    └── config
        └── parameters.yaml.gohtml

Here are some template examples:

production/.env.gohtml:

# This file was generated by Gomplate from Vault secrets for production
{{- range $key, $value := (datasource "vault:///foo_bar/data/env").data }}
{{ $key }}={{ $value | quote }}
{{- end }}

staging/.env.gohtml:

# This file was generated by Gomplate from Vault secrets for staging
{{- range $key, $value := (datasource (print "vault:///foo_bar/data/env" (has .Env "STAGE" | ternary (printf "-%s" (getenv "STAGE")) ""))).data }}
{{ $key }}={{ $value | quote }}
{{- end }}

Note the STAGE environnement variable usage, allowing to switch from env to env-${STAGE} vault secret, calling make secrets@staging STAGE=foo.

production/config/parameters.yaml.gohtml:

# This file was generated by Gomplate from Vault secrets for production
parameters:
{{ (datasource "vault:///foo_bar/data/parameters").data | toYAML | indent 4 -}}

Note

Note that the path to the secret will slightly differ from what the Vault server will display:
if the path is MyApp/production/env on the Vault server, it will become MyApp/data/production/env in the template

See Go Template syntax for more info.

Warning

Make sure to include the secrets directory into your deliveries, using the release_add entry.

Https

In order for https to work properly, you must:

MacOS

  1. ensure elao ca certificate has been added to your local keychain (one time for all projects)

    $ sudo security add-trusted-cert -d -r trustRoot -k "/Library/Keychains/System.keychain" .manala/certificates/ca.crt
  2. For firefox only, browse to about:config and ensure security.enterprise_roots.enabled value is set to true (one time for all projects)

  3. generate a project certificate (one time by project, remember to commit them right after)

    make provision.certificates

Ubuntu

  1. ensure elao ca certificate has been added to your local keychain (one time for all projects)

    $ sudo cp .manala/certificates/ca.crt /usr/local/share/ca-certificates
    $ sudo update-ca-certificates
  2. For Chrone and Firefox, add .manala/certificates/ca.crt certificate on Security and Privacy (one time for all projects).

    • On Chrome, chrome://settings/certificates > Authorities, import .manala/certificates/ca.crt.
    • On Firefox, about:preferences#privacy > View Certificate... > Authorities, import .manala/certificates/ca.crt.
  3. generate a project certificate (one time by project, remember to commit them right after)

    make provision.certificates

Extras

Check the following sections for extra gems included with this recipe.

Symfony

IDE configuration

As of Symfony >= 6.1, a SYMFONY_IDE env var allows you to configure your IDE once on your machine:

PhpStorm:

printf "SYMFONY_IDE=\"phpstorm://open?file=%%f&line=%%l\"" >> ~/.zshrc

Sublime:

printf "SYMFONY_IDE=\"subl://open?url=file://%%f&line=%%l\"" >> ~/.zshrc

However, since the paths in the container differs from the ones on your machine, you need an extra mapping per project. Hopefully, the Manala recipe will detect your host SYMFONY_IDE env var and forwards it to the container automatically with the proper paths mapping.

As of Symfony < 6.1, you can configure the same behavior with:

framework:
    ide: '%env(string:default::SYMFONY_IDE)%'

Caveats

  • MySQL 5.7 docker images are not available on arm64, that's why amd64 architecture is forced. Expect rare speed and stability issues.
  • Nginx official debian packages for stretch are not available on arm64, that's why debian packages are used in version 1.10. Expect rare configuration issues.
  • OpenSSL debian packages for buster are broken on arm64, that's why bullseye ones are used. Expect rare behavior issues.
  • Firefox is blocking some ports like 10080. Try to avoid them.