vmihailenco / fabdeploy Goto Github PK

• https://bitbucket.org/vmihailenco/fabdeploy
• https://github.com/vmihailenco/fabdeploy

There is full working example: https://github.com/vmihailenco/fabdeploy-example.

Create fabconf.py:

from fabdeploy.api import DefaultConf

class BaseConf(DefaultConf):
    django_dir = 'project_name'

class StagingConf(BaseConf):
    address = '[email protected]'

class ProdConf(BaseConf):
    address = '[email protected]'

Create fabfile.py:

from fabdeploy import monkey; monkey.patch_all()
from fabric.api import *
from fabdeploy.api import *; setup_fabdeploy()


@task
def user_create():
    users.create.run()
    ssh.push_key.run(pub_key_file='~/.ssh/id_rsa.pub')

@task
def deploy():
    pass

Fabdeploy uses two system (linux) users:

• sudo_user to perform tasks that require sudo right (root by default).
• user for other tasks (SSH user by default).

In Ubuntu root user is disabled by default. You can create special fabdeploy user using following command:

fab fabd.default_conf:address=user@host,sudo_user=user fabd.create_user

Then you should tell fabdeploy to use new sudo_user:

class ProdConf(BaseConf):
    sudo_user = 'fabdeploy'

List of available tasks:

fab --list

List of available variables:

fab fabd.debug

This is useful to test configuration:

$ fab fabd.conf:prod fabd.debug:django_path
/home/prj/src/prj

or:

$ fab fabd.conf:prod fabd.debug:cpu_count
2

or:

$ fab fabd.conf:prod fabd.debug:current_time
2011.11.27-13.40

To deploy project you may use:

$ fab fabd.conf:staging deploy
$ fab fabd.conf:prod deploy

fabconf.py:

from fabdeploy.api import DefaultConf

class ProdConf(DefaultConf):
    my_task__log_path = '/var/log/my_task'

fabfile.py:

from fabdeploy.api import Task

class MyTask(Task):
    def do(self):
        print self.conf.log_path

my_task = MyTask()

Output:

$ fab fabd.conf:prod my_task
/var/log/my_task

You can also temporarily set log path:

$ fab fabd.conf:prod my_task:log_path='/var'
/var

This works for all variables and all tasks.

fabconf.py:

from fabdeploy.api import DefaultConf

class ProdConf(DefaultConf):
    # default DB
    db_name = 'name1'
    db_user = 'user1'
    db_password = 'pass1'
    # logging DB
    loggingdb__db_name = 'name2'
    loggingdb__db_user = 'user2'
    loggingdb__db_password = 'pass2'

fabfile.py:

from fabdeploy import postgres

@task
def dump_db():
    postgres.dump.run()  # dump default DB
    postgres.dump.run(_namespace='loggingdb__')  # dump logging DB

Fabdeploy is written to be highly configurable. For example, there is built-in tar task, which by default packs whole project, uploads it to server and unpacks it there.

But you can freely use it to upload custom dirs:

from fabdeploy import tar

@task
def push_static():
    tar.push.run(
        src_dir=os.path.join(env.conf.django_ldir, 'static'),
        target_dir=posixpath.join(env.conf.django_dir, 'static'))

fabconf.py:

from fabdeploy import api
from fabdeploy.api import DefaultConf

class DevConf(DefaultConf):
    address = 'user@localhost'
    db = getattr(fabdeploy, 'mysql')

class ProdConf(DefaultConf):
    address = 'user@localhost'
    db = getattr(fabdeploy, 'postgres')

fabfile.py:

@task
def execute():
    print env.conf.db.execute

There are some conventions how to configure fabdeploy:

• You should extend DefaultConf:

  from fabdeploy.api import DefaultConf
  
  class BaseConf(DefaultConf):
      pass
  

• Each value can contain Python formatting:

  class BaseConf(DefaultConf):
      supervisor__log_dir = '%(var_dir)s/log/supervisor'
  

• Remote pathes should have posfix _path. You can and should use task fabd.mkdirs to create all remote dirs with one command. It will look like this:

  $ fab fabd.conf:staging_conf fabd.mkdirs
  mkdir --parents /path/to/dir1 /path/to/dir2 /path/to/dir3
  

• Remote dirs (e.g. var) have postfix _dir.

• Local pathes have postfix _lpath. Local dirs have postfix _ldir. This is similar to Fabric cd and lcd tasks.

• Dirs (postfix _dir and _ldir) and pathes (postfix _path and _lpath) can be Python lists. These lists will be passed to os.path.join() or posixpath.join(). Previous example can look like this:

  from fabdeploy.api import DefaultConf
  
  class BaseConf(DefaultConf):
      supervisor__log_dir = ['%(var_dir)s', 'log', 'supervisor']
  

• Function can be decorated with conf decorator. For example, current_time task looks like this:

  from fabdeploy.api import DefaultConf
  
  class BaseConf(DefaultConf):
      @conf
      def current_time(self):
          return datetime.datetime.utcnow().strftime(self.time_format)
  

  You can use it in your task like this:

  from fabdeploy.api import Task
  
  class MyTask(Task):
      def do(self):
          puts(self.conf.current_time)
  

• You can configure each task individually:

  class BaseConf(DefaultConf):
      postgres__db_name = 'postgresql_db'  # module=postres
      mysql__db_name = 'mysql_db'          # module=mysql
      mysql__create_db__db_user = 'root'   # module=mysql, task=create_db
  

Configuration is stored in task instance variable self.conf. Each task has its own copy of configuration. Configuration variables are searched in following places:

• task keyword argument var (fab task:foo=bar);
• task instance method var() decorated with @conf();
• key var in env.conf, which is populated by fabd.conf task;
• ask user to provide variable var using fabric prompt.

Global configuration is stored in env.conf.

Your task is class-based fabric class except fabdeploy manages configuration for you:

from fabric.api import puts
from fabdeploy.api import Task, conf

class MessagePrinter(Task):
    @conf
    def message(self):
        if 'message' in self.conf:
            return self.conf.message
        return 'Hi!'

    def do(self):
        if self.conf.secret == '123':
            puts(self.conf.message)
        else:
            puts('huh?')

message_printer = MessagePrinter()

Then you can run this task like this:

$ fab message_printer
> secret = 123
Hi!
$ fab message_printer:message='Hello world!'
> secret = 123
Hello world!

Typical fabfile may look like this:

from fabdeploy import monkey; monkey.patch_all()
from fabric.api import *
from fabdeploy.api import *; setup_fabdeploy()

@task
def install():
    users.create.run()
    ssh.push_key.run(pub_key_file='~/.ssh/id_rsa.pub')

    system.setup_backports.run()
    system.install_common_software.run()

    with settings(warn_only=True):
        postgres.create_role.run()
        postgres.create_db.run()
        postgres.grant.run()

    nginx.install.run()

    for app in ['supervisor']:
        pip.install.run(app=app)


@task
def setup():
    fabd.mkdirs.run()

    nginx.push_gunicorn_config.run()
    nginx.restart.run()

    supervisor.d()


@task
def deploy():
    fabd.mkdirs.run()
    release.create.run()

    postgres.dump.run()

    git.init.run()
    git.push.run()

    supervisor.push_configs.run()
    django.push_settings.run()
    gunicorn.push_config.run()

    virtualenv.create.run()
    virtualenv.pip_install_req.run()
    virtualenv.pip_install.run(app='gunicorn')
    virtualenv.make_relocatable.run()

    django.syncdb.run()
    django.migrate.run()
    django.collectstatic.run()

    release.activate.run()

    supervisor.update.run()
    supervisor.restart_program.run(program='celeryd')
    gunicorn.reload_with_supervisor.run()