$ pip install boss-cli==1.0.2
Check the configuration page.
Custom scripts are scripts/commands that could be defined directly in the config file without having to write any line of python in the fabfile.py
. They're similar to the npm scripts, if you're familiar with them.
You can define the custom scripts under the scripts
field in the boss.yml
.
For instance:
# boss.yml
stages:
prod:
host: your-server.com
public_url: 'https://your-server.com'
branch: master
scripts:
hello: 'echo "Hello World!"'
build: npm run build
logs: pm2 logs
Boss comes out of the box with a task run
which you can use to run these scripts on the remote server like this:
$ fab prod run:hello
$ fab prod run:build
$ fab prod run:logs
This is a generic deployment preset, where the remote host also contains the project source code and the git repository. The deploy task would synchronize the remote with the latest changes of the provided branch from the origin. It then builds the project and restarts the service if needed.
This is general and language agnostic so it can be used for deploying any kind of project. You just need to specify the relevant build
script to build your project in the remote and if it requires service restart then you'll need to define a reload
script as well.
You'll need to set the deployment preset as remote-source
in your configuration.
deployment:
preset: remote-source
Your boss.yml
file for remote source deployment would look similar to this:
project_name: my-app
project_description: 'My App'
repository_url: 'https://github.com/username/repository'
branch_url: '{repository_url}/tree/{branch}'
user: deploy_user
deployment:
preset: remote-source
base_dir: /source/my-app
stages:
prod:
host: your-server.com
public_url: 'https://your-server.com'
branch: master
scripts:
install: 'npm install'
build: 'npm run build'
start: 'pm2 start dist/myapp.js'
stop: 'pm2 stop dist/myapp.js'
reload: 'pm2 reload dist/myapp.js'
notifications:
slack:
enabled: true
endpoint: ${BOSS_SLACK_ENDPOINT}
The above configuration is specific to a Node.js project environment, but you can also deploy projects built with other languages like PHP, Python, Java etc. All you need to do is change the scripts install
, build
, reload
.
You can check the available tasks for remote-source
preset with fab --list
.
➜ fab --list
Available commands:
build Build the application.
check Check the current remote branch and the last commit.
deploy Deploy to remote source.
prod Configures the prod server environment.
logs Tail the logs.
restart Restart the service.
run Run a custom script.
status Check the status of the service.
stop Stop the service.
sync Sync the changes on the branch with the remote (origin).
Now to deploy the the application to the prod
server that you've configured in the stages
above. You can do:
➜ fab prod deploy
This would deploy the default branch master
in this case. You can also provide a specific branch to deploy, as follows:
➜ fab prod deploy:branch=my-branch
This deployment is useful for deploying the web apps (React, Angular, Vue etc) or static files to the remote server. This preset assumes the static files are served via a web server on the remote host eg: nginx, apache etc. Here, the source code is built locally and only the dist
or build
is uploaded and deployed to the server.
The deployment process is zero-downtime, just like capistrano.
You'll need to set the deployment preset as web
in your configuration.
deployment:
preset: web
Your boss.yml
file for web deployment would look similar to this:
project_name: my-app
project_description: 'My App'
repository_url: 'https://github.com/username/repository'
branch_url: '{repository_url}/tree/{branch}'
user: deploy_user
deployment:
preset: web
build_dir: build/ # The local build directory
base_dir: /app/deployment # The remote base directory for deployment.
stages:
prod:
host: your-server.com
public_url: 'https://your-server.com'
scripts:
install: 'npm install'
build: 'npm run build'
notifications:
slack:
enabled: true
endpoint: ${BOSS_SLACK_ENDPOINT}
The above configuration would work for any kind of web projects as long as it generates the build in static files (HTML, CSS, JS, media) that could be served via a web server.
You can define two scripts install
and build
in your boss.yml
, to install project dependencies and build the source respectively. For instance: if you've created your application using create-react-app
, you can set these to npm install
and npm run build
as shown in above config.
You also have to set the location of the output directory for the build
script as deployment.build_dir
. In our case, this would be the build/
directory.
You can check the available tasks for this preset with fab --list
.
➜ fab --list
Available commands:
build Build the code locally.
buildinfo Print the build information.
builds Display the build history.
deploy Zero-Downtime deployment for the web.
info Print the build information.
logs Tail the logs.
rollback Zero-Downtime deployment rollback for the web.
run Run a custom script.
setup Setup remote host for deployment.
prod Configures the prod server environment.
For the first time, you can configure the remote host for deployment using the setup
task.
➜ fab prod setup
This will create necessary files and directories on the remote under the provided base_dir
path. In our case the base directory will be /app/deployment
.
Once, the setup
task completes you should see a message like this:
Remote is setup and is ready for deployment.
Deployed build will point to /app/deployment/current.
For serving the latest build, please set your web server document root to /app/deployment/current.
Now you'll need to set your web server document root on the remote host to the current
symlink created under the base_dir
path. This symlink will point to the latest build when you deploy your app.
If you're using a web server like nginx, you can set the document root like this:
# Sample nginx Configuration.
server {
listen 80;
listen [::]:80;
# This is the symlink that points to the build that is deployed.
root /app/deployment/current;
index index.html;
...
}
You can use the deploy
task to deploy the app to the remote server.
Here, first the deploy
task would trigger the install
and build
scripts to build the project locally, after which the built directory configured in deployment.build_dir
would be tar-zipped and uploaded to the remote host via SSH.
So, to deploy current local source code to prod
server you should do the following:
➜ fab prod deploy
If you're using git
in your project, you need to make sure you did checkout
to the branch you want to deploy and is up to date. Like this,
# Checkout to the right branch and deploy
➜ git checkout master
➜ fab prod deploy
Node.js Project Deployment. More information, examples and documentation coming soon :).
Special Thanks to Shirish Shikhrakar for the logo.
All kinds of contributions are welcome. Read our contributing guide to learn about our development process, how to propose PRs, report bugs and suggest improvements.
Check the CHANGELOG for full release history.
Licensed under The MIT License.