capistrano-archive

This plugin adds a new scm to Capistrano 3 allowing deployment from an archive stored on the target server (s).

• Upload an archive (tar.bz2 as default) to target
• Create release from the archive
• Can use a file from the archive to know the deployed revision (./REVISION as default)
• Workflow between deploy stages copying the local archive on stage directory

Put this in the Gemfile

    gem 'capistrano-archive'

Then run

    bundle install

Add theses lines to your Capfile

# Include and use capistrano archive 
require "capistrano/archive"
install_plugin Capistrano::Archive::SCM

First, you need to have a build system storing artefact release on the server used to deploy.

To specify the archive location, you can set the :archive_release_path variable. As default, the archive is supposed to be stored in the current directory.

The archive name is built as :application_:branch but you can change that with the :archive_name variable.

The tar is compressed with bzip as default but you can change that with :archive_extension and :archive_tar_options variables.

As default the archive is sent in the :tmp_dir but you can change the destination using the :archive_remote_path variable.

If you have multiple servers in same stage, you can upload the archive on some servers only using the :archive_roles variable.

Using a shared storage for theses servers, you can upload the archive one time only in a shared location and use it on each server. For example like that :

  set :archive_roles, :upload
  set :archive_remote_path, "/nas/..."

If you want have workflow between stage, like force deploy on staging before production, you can set the :archive_workflow_previous variable in your stages config file.

For example, in a staging.rb file, set the variable like that

  set :archive_workflow_previous, "build"  

And the archive location will be :archive_releases_path/build when you deploy the staging stage. After the deploy, the archive file will be copied in the :archive_workflow_next, as default the current stage.

Do the same in a production.rb file

  set :archive_workflow_previous, "staging"

And the archive location will be :archive_releases_path/staging when you deploy the production stage. So you'll have to deploy on staging before production.

If you want turn off the workflow process, set the variable :archive_workflow to false.

Capistrano log the revision info, the sha1 commit, in the revisions log file and in the file REVISION. Since we are deploying from an archive, if we want to have the same info, it must already be present in the archive. The most easy is to store the commit when build the archive in the REVISION file.

As default, the plugin can read the REVISION file from the archive.

If you want to store the info in another file, you can set the file with the :archive_revision_file variable. If you have to manipulate the output (grep, awk, etc) you can change the :archive_revision_command

If you want turn off the revision process, set the variable :archive_revision to false.

If you need to change default values, you can set the following configuration variables in config/deploy.rb or wherever relevant:

  set :archive_releases_path, "./"
  set :archive_name, "#{fetch(:application)}_#{fetch(:branch)}"
  set :archive_extension, 'tar.bz2'
  set :archive_tar_options, 'xjf'
  set :archive_workflow, true
  set :archive_workflow_previous, ""
  set :archive_workflow_next,"#{fetch(:stage)}"
  set :archive_remote_path, "#{fetch(:tmp_dir)}"
  set :archive_revision, true
  set :archive_revision_file, './REVISION'
  set :archive_revision_command, 'cat'

Please feel free to open an issue or a pull request, if you want to help.

Thanks to everyone who has contributed already.

This project is supported by PrestaConcept

Released under the MIT License