Skip to content

Latest commit

 

History

History
64 lines (44 loc) · 2.73 KB

README.md

File metadata and controls

64 lines (44 loc) · 2.73 KB
Raw

deploy.sh

Deploy your project(s) using shell scripts.

Requirements

This deployment strategy assumes:

  • a server to build your project(s)
  • bash is installed on every involved servers (build and target servers)
  • project(s) versioned under git
  • the build server can access target servers using ssh and rsync

How it works?

This script is composed of mainly 2 commands: build and release.
The first one is meant to create an archive file for a given revision.
It should be immutable in a way that if you call this command twice for the same revision, it should produce the exact same content: an archive.

The latter command deploy an archive produced by the build command to a determined target server(s).

The build command

deploy build [options] <revision>

The build process

  1. Refresh the local repository. The script maintains a repository stored within the build server. If you have specified a remote repository url, this steps will refresh the local repository
  2. Create a workspace in the build server's temporary directory.
  3. Extract git content of the specified revision using git archive
  4. Generate a .REVISION file in the root directory of the workspace. This file will contain the commit sha of the current revision.
  5. Call a user defined hook name "build". This is where you can add some specific action to do during the build step.
  6. Archive the workspace in a tar.bz2 file.

For more details about this command, use the --help option or visit the build documentation

The release command

deploy release [options] <config-file> <archive-file> [<server-name> ...]

config-file: a configuration file containing every details needed to deploy to servers. TODO: explain the config file content
archive-file: the archive file produced by the build command.
server-name: optional. Name of one or more servers to deploy to. It filters the server list provided by the config-file and allows to deploy to less servers than in a normal process.

The release process

  1. Test ssh connection to specified servers
  2. Ensure directory structure exists within each servers, creating it if not. TODO: explain the directory structure
  3. Ensure defined shared items exists in the shared folder. Call hook "create_shared_links" if an item does not exists and check again.
  4. Send archive to every servers
  5. Extract archive within every servers and rename the folder with the current date
  6. Link release with shared items defined in the config file
  7. Activate the release. TODO: explain what it means
  8. Clean old releases.

Use the --help option for more details about this command.

Advanced features