Web   ·   Wiki   ·   Activities   ·   Blog   ·   Lists   ·   Chat   ·   Meeting   ·   Bugs   ·   Git   ·   Translate   ·   Archive   ·   People   ·   Donate
summaryrefslogtreecommitdiffstats
path: root/studio/static/doc/flask-docs/_sources/patterns/fabric.txt
blob: b02ad27720ae15a28e909f4587afc689619e0af6 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
.. _fabric-deployment:

Deploying with Fabric
=====================

`Fabric`_ is a tool for Python similar to Makefiles but with the ability
to execute commands on a remote server.  In combination with a properly
set up Python package (:ref:`larger-applications`) and a good concept for
configurations (:ref:`config`) it is very easy to deploy Flask
applications to external servers.

Before we get started, here a quick checklist of things we have to ensure
upfront:

-   Fabric 1.0 has to be installed locally.  This tutorial assumes the
    latest version of Fabric.
-   The application already has to be a package and requires a working
    `setup.py` file (:ref:`distribute-deployment`).
-   In the following example we are using `mod_wsgi` for the remote
    servers.  You can of course use your own favourite server there, but
    for this example we chose Apache + `mod_wsgi` because it's very easy
    to setup and has a simple way to reload applications without root
    access.

Creating the first Fabfile
--------------------------

A fabfile is what controls what Fabric executes.  It is named `fabfile.py`
and executed by the `fab` command.  All the functions defined in that file
will show up as `fab` subcommands.  They are executed on one or more
hosts.  These hosts can be defined either in the fabfile or on the command
line.  In this case we will add them to the fabfile.

This is a basic first example that has the ability to upload the current
sourcecode to the server and install it into a pre-existing
virtual environment::

    from fabric.api import *

    # the user to use for the remote commands
    env.user = 'appuser'
    # the servers where the commands are executed
    env.hosts = ['server1.example.com', 'server2.example.com']

    def pack():
        # create a new source distribution as tarball
        local('python setup.py sdist --formats=gztar', capture=False)

    def deploy():
        # figure out the release name and version
        dist = local('python setup.py --fullname', capture=True).strip()
        # upload the source tarball to the temporary folder on the server
        put('dist/%s.tar.gz' % dist, '/tmp/yourapplication.tar.gz')
        # create a place where we can unzip the tarball, then enter
        # that directory and unzip it
        run('mkdir /tmp/yourapplication')
        with cd('/tmp/yourapplication'):
            run('tar xzf /tmp/yourapplication.tar.gz')
            # now setup the package with our virtual environment's
            # python interpreter
            run('/var/www/yourapplication/env/bin/python setup.py install')
        # now that all is set up, delete the folder again
        run('rm -rf /tmp/yourapplication /tmp/yourapplication.tar.gz')
        # and finally touch the .wsgi file so that mod_wsgi triggers
        # a reload of the application
        run('touch /var/www/yourapplication.wsgi')

The example above is well documented and should be straightforward.  Here
a recap of the most common commands fabric provides:

-   `run` - executes a command on a remote server
-   `local` - executes a command on the local machine
-   `put` - uploads a file to the remote server
-   `cd` - changes the directory on the serverside.  This has to be used
    in combination with the `with` statement.

Running Fabfiles
----------------

Now how do you execute that fabfile?  You use the `fab` command.  To
deploy the current version of the code on the remote server you would use
this command::

    $ fab pack deploy

However this requires that our server already has the
``/var/www/yourapplication`` folder created and
``/var/www/yourapplication/env`` to be a virtual environment.  Furthermore
are we not creating the configuration or `.wsgi` file on the server.  So
how do we bootstrap a new server into our infrastructure?

This now depends on the number of servers we want to set up.  If we just
have one application server (which the majority of applications will
have), creating a command in the fabfile for this is overkill.  But
obviously you can do that.  In that case you would probably call it
`setup` or `bootstrap` and then pass the servername explicitly on the
command line::

    $ fab -H newserver.example.com bootstrap

To setup a new server you would roughly do these steps:

1.  Create the directory structure in ``/var/www``::

        $ mkdir /var/www/yourapplication
        $ cd /var/www/yourapplication
        $ virtualenv --distribute env

2.  Upload a new `application.wsgi` file to the server and the
    configuration file for the application (eg: `application.cfg`)

3.  Create a new Apache config for `yourapplication` and activate it.
    Make sure to activate watching for changes of the `.wsgi` file so
    that we can automatically reload the application by touching it.
    (See :ref:`mod_wsgi-deployment` for more information)

So now the question is, where do the `application.wsgi` and
`application.cfg` files come from?

The WSGI File
-------------

The WSGI file has to import the application and also to set an environment
variable so that the application knows where to look for the config.  This
is a short example that does exactly that::

    import os
    os.environ['YOURAPPLICATION_CONFIG'] = '/var/www/yourapplication/application.cfg'
    from yourapplication import app

The application itself then has to initialize itself like this to look for
the config at that environment variable::

    app = Flask(__name__)
    app.config.from_object('yourapplication.default_config')
    app.config.from_envvar('YOURAPPLICATION_CONFIG')

This approach is explained in detail in the :ref:`config` section of the
documentation.

The Configuration File
----------------------

Now as mentioned above, the application will find the correct
configuration file by looking up the `YOURAPPLICATION_CONFIG` environment
variable.  So we have to put the configuration in a place where the
application will able to find it.  Configuration files have the unfriendly
quality of being different on all computers, so you do not version them
usually.

A popular approach is to store configuration files for different servers
in a separate version control repository and check them out on all
servers.  Then symlink the file that is active for the server into the
location where it's expected (eg: ``/var/www/yourapplication``).

Either way, in our case here we only expect one or two servers and we can
upload them ahead of time by hand.

First Deployment
----------------

Now we can do our first deployment.  We have set up the servers so that
they have their virtual environments and activated apache configs.  Now we
can pack up the application and deploy it::

    $ fab pack deploy

Fabric will now connect to all servers and run the commands as written
down in the fabfile.  First it will execute pack so that we have our
tarball ready and then it will execute deploy and upload the source code
to all servers and install it there.  Thanks to the `setup.py` file we
will automatically pull in the required libraries into our virtual
environment.

Next Steps
----------

From that point onwards there is so much that can be done to make
deployment actually fun:

-   Create a `bootstrap` command that initializes new servers.  It could
    initialize a new virtual environment, setup apache appropriately etc.
-   Put configuration files into a separate version control repository
    and symlink the active configs into place.
-   You could also put your application code into a repository and check
    out the latest version on the server and then install.  That way you
    can also easily go back to older versions.
-   hook in testing functionality so that you can deploy to an external
    server and run the testsuite.  

Working with Fabric is fun and you will notice that it's quite magical to
type ``fab deploy`` and see your application being deployed automatically
to one or more remote servers.


.. _Fabric: http://fabfile.org/