pkgr
Goal
Make debian or rpm packages out of any app, including init script, crons, logrotate, etc. Excellent way to distribute apps or command line tools without complicated installation instructions.
Hosted service available at Packager.io. Free for OpenSource apps.
Officially supported languages
- Ruby
- NodeJS
- Python
- Go
In beta:
- PHP, using the following buildpack:
https://github.com/pkgr/heroku-buildpack-php#buildcurl
You can also point to other buildpacks (doc). They may just work.
Supported distributions (64bits only)
- Ubuntu 22.04 ("jammy")
- Ubuntu 20.04 ("focal")
- Ubuntu 18.04 ("bionic")
- Ubuntu 16.04 ("xenial")
- Ubuntu 14.04 ("trusty")
- Ubuntu 12.04 ("precise")
- Debian 11 ("bullseye")
- Debian 10 ("buster")
- Debian 9 ("stretch")
- Debian 8 ("jessie")
- Debian 7 ("wheezy")
- RHEL/CentOS 8
- RHEL/CentOS 7
- RHEL/CentOS 6
- Suse Linux Enterprise Server 12
- Suse Linux Enterprise Server 11
- Amazon Linux AMI 2015
- Amazon Linux AMI 2014
Examples
See Packager.io for examples of apps packaged with pkgr
(Gitlab, OpenProject, Discourse, etc.).
Installation
Docker images are available for the most recent distributions. If you don't
have docker or are interested in older distributions, you can also install
pkgr
as a ruby gem:
gem install pkgr
Usage
To package your app, execute pkgr
against your app's repository:
# if using the ruby gem, you need to run it under the distribution you want to package for:
pkgr package path/to/app/repo
# if using a docker image, you can run it from any host OS:
docker run --rm -it -v $(pwd):/app -v /tmp/cache:/cache pkgr/ubuntu:20.04
docker run --rm -it -v $(pwd):/app -v /tmp/cache:/cache pkgr/el:8
docker run --rm -it -v $(pwd):/app -v /tmp/cache:/cache pkgr/debian:10
The resulting .deb
or .rpm
package will be in your current working directory.
Full command line options are given below:
$ pkgr help package
Usage:
pkgr package TARBALL|DIRECTORY
Options:
[--buildpack=BUILDPACK] # Custom buildpack to use
[--buildpack-list=BUILDPACK_LIST] # Specify a file containing a list of buildpacks to use (--buildpack takes precedence if given)
[--changelog=CHANGELOG] # Changelog
[--maintainer=MAINTAINER] # Maintainer
[--vendor=VENDOR] # Package vendor
[--architecture=ARCHITECTURE] # Target architecture for the package
# Default: x86_64
[--runner=RUNNER] # Force a specific runner (e.g. upstart-1.5, sysv-lsb-1.3)
[--logrotate-frequency=FREQUENCY] # Set logrotate frequency
# Default: daily
# Possible values: daily, weekly, monthly, yearly
[--logrotate-backlog=BACKLOG] # Set logrotate backlog
# Default: 14
[--homepage=HOMEPAGE] # Project homepage (e.g. "https://pkgr.example.org")
[--home=HOME] # Project home (e.g. "/usr/share/PACKAGE_HOME")
[--description=DESCRIPTION] # Project description
[--category=CATEGORY] # Category this package belongs to
# Default: none
[--version=VERSION] # Package version (if git directory given, it will use the latest git tag available)
[--iteration=ITERATION] # Package iteration (you should keep the default here)
# Default: 20141015024539
[--license=LICENSE] # The license of your package (see <https://www.debian.org/doc/packaging-manuals/copyright-format/1.0/#license-short-name>)
[--user=USER] # User to run the app under (defaults to your app name)
[--group=GROUP] # Group to run the app under (defaults to your app name)
[--compile-cache-dir=COMPILE_CACHE_DIR] # Where to store the files cached between packaging runs. Path will be resolved from the temporary code repository folder, so use absolute paths if needed.
[--before-precompile=BEFORE_PRECOMPILE] # Provide a script to run just before the buildpack compilation, on the build machine. Path will be resolved from the temporary code repository folder, so use absolute paths if needed.
[--after-precompile=AFTER_PRECOMPILE] # Provide a script to run just after the buildpack compilation, on the build machine. Path will be resolved from the temporary code repository folder, so use absolute paths if needed.
[--before-install=BEFORE_INSTALL] # Provide a script to run just before a package gets installated or updated, on the target machine.
[--after-install=AFTER_INSTALL] # Provide a script to run just after a package gets installated or updated, on the target machine.
[--before-remove=BEFORE_REMOVE] # Provide a script to run just before a package gets uninstallated, on the target machine.
[--after-remove=AFTER_REMOVE] # Provide a script to run just after a package gets uninstallated, on the target machine.
[--dependencies=one two three] # Specific system dependencies that you want to install with the package
[--build-dependencies=one two three] # Specific system dependencies that must be present before building
[--host=HOST] # Remote host to build on (default: local machine)
[--auto], [--no-auto] # Automatically attempt to install missing dependencies
[--clean], [--no-clean] # Automatically clean up temporary dirs
# Default: true
[--edge], [--no-edge] # Always use the latest version of the buildpack if already installed
# Default: true
[--env=one two three] # Specify environment variables for buildpack (--env "CURL_TIMEOUT=2" "BUNDLE_WITHOUT=development test")
[--force-os=FORCE_OS] # Force a specific distribution to build for (e.g. --force-os "ubuntu-12.04"). This may result in a broken package.
[--store-cache], [--no-store-cache] # Output a tarball of the cache in the current directory (name: cache.tar.gz)
[--verify], [--no-verify] # Verifies output package
# Default: true
[--data-dir=DATA_DIR] # Custom path to data directory. Can be used for overriding default templates, hooks(pre-, post- scripts), configs (buildpacks, distro dependencies), environments, etc.
# Default: ./pkgr/data
[--directories=DIRECTORIES] # Recursively mark a directory as being owned by the package
[--verbose], [--no-verbose] # Run verbosely
[--debug], [--no-debug] # Run very verbosely
[--name=NAME] # Application name (if directory given, it will default to the directory name)
[--buildpacks-cache-dir=BUILDPACKS_CACHE_DIR] # Directory where to store the buildpacks
# Default: /home/vagrant/.pkgr/buildpacks
Why?
Tools such as Capistrano are great for deploying applications, but the deployment recipe can quickly become a mess, and scaling the deployment to more than a few servers can prove to be difficult. Plus, if you're already using automation tools such as Puppet to configure your servers, you have to run two different processes to configure your infrastructure.
pkgr
builds on top of the Heroku tools to provide you with an easy way to package you app as a debian package. The great advantage is that once you've built it and you tested that it
works once, you can deploy on any number of servers at any time and you're
sure that it will just work. Then, you can upgrade/downgrade or uninstall the whole application in one command.
Finally, it's a great way to share your open source software with your users and clients. Much easier than asking them to install all the dependencies manually! I'm in the process of making sure pkgr
is feature complete by trying to package as many successful open-source projects as I can. Don't hesitate to test it on your app and report your findings!
What this does
- Uses Heroku buildpacks to embed all the dependencies related to your application runtime within the debian package. For a Rails app for instance, this means that
pkgr
will embed the specific ruby runtime you asked for, along with all the gems specified in your Gemfile. However, all other dependencies you may need must be specified as additional system dependencies (see Usage). This avoids the 'packaging-the-world' approach used by other tools such as omnibus (with the pros and cons that come with it), but it still allows you to use the latest and greatest libraries for your language of choice. See this blog post for more background.
-
Gives you a nice executable, which closely replicates the Heroku toolbelt utility. For instance, assuming you're packaging an app called
my-app
, you can do the following:my-app config:set VAR=value my-app config:get VAR my-app run [procfile process] # e.g. my-app run rake db:migrate; my-app run console; etc. my-app run [arbitrary process] # e.g. my-app run ruby -v; my-app run bundle install; etc. my-app scale web=1 worker=1 my-app logs [--tail] ...
-
Your app will reside in
/opt/app-name
. -
You'll also get upstart, systemd, or sysvinit initialization scripts (depending on your distribution) that you can use directly:
service my-app start/stop/restart/status
Note: init scripts are generated only after you've done a
scale
command for the process type, e.g.my-app scale web=1
. -
Logs will be stored in
/var/log/app-name/
, with a proper logrotate config automatically added. For systemd-based distributions, you will find the logs in the systemd journal. -
Config files can be added in
/etc/app-name/
Requirements
-
You must have a Procfile.
-
Your application should be Heroku compatible, meaning you should be able to set your main app's configuration via environment variables.
Troubleshooting
If you're on older versions of Debian, you may need to append /var/lib/gems/1.9.1/bin
to your PATH to "see" the pkgr
command:
export PATH="$PATH:/var/lib/gems/1.9.1/bin"
If you get the following error ERROR: While executing gem ... (ArgumentError) invalid byte sequence in US-ASCII
while trying to install pkgr
, try setting a proper locale, and then retry:
sudo locale-gen en_US.UTF-8
export LANG=en_US.UTF-8
export LC_ALL=en_US.UTF-8
sudo gem install pkgr
Looking for the init script? It is created the first time you run this command
sudo my-app scale web=1 worker=1
Issue getting nokogiri to compile? Try the following based on this comment:
bundle config --local build.nokogiri "--use-system-libraries --with-xml2-include=/usr/include/libxml2"
Authors
- Cyril Rohr [email protected] - http://crohr.me, https://packager.io
Copyright
See LICENSE (MIT)