Skip to content

A shell script for deploying web apps from Git repositories.

License

Notifications You must be signed in to change notification settings

ediblemanager/deploy_from_git

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Overview
========

deploy_from_git.sh is a shell script for deploying web apps from Git
repositories. It should be fairly generic. It's been used to deploy
PHP and Ruby on Rails apps.

It keeps a configurable (defaults to 2) number of deploys around. The
current one is signified by the "current" symlink. Unlike some other
deployment mechanisms it doesn't result in a copy of the repository
metatdata being deployed. It uses git-archive to exract just the files
needed.

Use
===

    $ deploy_from_git.sh -h
    Usage: deploy_from_git.sh [-t <target>] <SHA1 | HEAD | dirty>

It takes one or two arguments. An optional target argument and the commit
you wish to deploy. You can specify either HEAD, the SHA1 of the commit
you wish to deploy (currently it only uses the first 8) or dirty.

Dirty is HEAD + uncommitted changes, useful for when your doing testing.

e.g Default use from within the app repository

    $ deploy_from_git.sh aa3a5583a1cc31205390080b5309e20c48606fee


Configuration
=============

It needs some configuration before use however. It is configured on a
per repository basis through simple text files.

For each repository you wish to deploy from you should create a file in
~/.config/deploy_from_git/<repository>.config (creating the deploy_from_git
if required). <repository> is simply the name of the directory that contains
the repository you are deploying from.

The config file itself should have two values.

    HOST="host.domain"
    REMOTE_APP_DIR="/path/to/app"

Self explanatory, but HOST is the name of the machine your deploying to
and REMOTE_APP_DIR is the path to the location of the app on the server.

Alternative deploy targets
--------------------------

By default the script will use ~/.config/deploy_from_git/<repository>.config
as the config file.

You can however provide alternate target config files, e.g

    ~/.config/deploy_from_git/<repository>.testing.config

Here we provide an alternate "testing" target. To access it, call the script
as (argument order is important)

    $ deploy_from_git.sh -t testing aa3a5583a1cc31205390080b5309e20c48606fee

You can have as many of these as you want as long you follow the above
naming scheme.

Excludes
--------

You can optionally create an excludes file to prevent certain
files/directories from being uploaded. If this file exists it is passed to
the rsync command.

It should be called <repository>.excludes and stored under
~/.config/deploy_from_git/

You can also have per target excludes like with the deploy targets. Just
name your excludes file to match e.g

    <repository>.<target>.excludes

If you specify a target but don't have an excludes file to match it will
use the main excludes file if it exists.

See the INCLUDE/EXCLUDE PATTERN RULES section of the rsync(1) man page
for details of what this file can take.

Number of releases
------------------

You can optionally specify the number of releases to have at any one time
on the server. It defaults to 2 if none or less than 2 is specified.

To set it, add a line to ~/.config/deploy_from_git/<repository>.config like

    NR_RELEASES=4

Post Run Script
---------------

You can specify a file that contains a list of commands to be run by the
script at the end of the deployment. This file shouldn't be executable or
contain a hash-bang line.

This file should be named <repository>.post or <repository>.<target>.post
and stored under ~/.config/deploy_from_git/

Unlike with the excludes. the post file will only be read for its specific
target. i.e if you specify a target of test and there is no .test.post file,
but there is a main .post file, this .post will _not_ be read.

Example
=======

So taking my-app as an example.

    $ pwd
    /home/andrew/programming/projects/my-app

So I would create a config file called

    ~/.config/deploy_from_git/my-app.config

That contains something like

    HOST="host.example.com"
    REMOTE_APP_DIR="/var/www/sites/my-app"

Then to do a deploy, say from the latest commit with a SHA1 of
b63e72515f8b1dcafad59079bc27fe1c9f61cbb2

    $ deploy_from_git.sh HEAD

This will be truncated down to the first 8 characters (b63e7251)

You'll see a bunch of messages scroll by, prefixed with either; [Local],
[Remote} or [Error]. If you get an [Error] pay attention, something went
wrong and the deploy likely failed.


If I wanted to exclude some files/directories from being uploaded I would
create a file

    ~/.config/deploy_from_git/my-app.excludes

containing e.g

    COPYING
    README
    docs/
    sql/

About

A shell script for deploying web apps from Git repositories.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Shell 100.0%