ops2deb

Are you tired of checking if your favorite devops tools are up-to-date? Are you using a debian based GNU/Linux distribution? ops2deb is designed to generate Debian packages for common devops tools such as kubectl, kustomize, helm, ..., but can be used to package any portable application. In short, it consumes a configuration file and outputs .deb packages. ops2deb can also track new releases of upstream applications and automatically bump application versions in its configuration file.

• Installation
  • With <a href="https://docs.wakemeops.com" rel="nofollow">wakemeops</a>
  • With <a href="https://github.com/pipxproject/pipx">pipx</a>
• Dependencies
• Getting started
• Usage examples
  • Packaging kubectl
  • Creating a metapackage
  • Packaging ops2deb with ops2deb
  • Building packages for multiple architectures at once
  • Using environment variables
• Configuration file
• Development
• Important notes
• Migration guides
  • Migrating to v1
  • Breaking changes in v2

Installation

With wakemeops

sudo apt-get install ops2deb

With pipx

pipx install ops2deb

Dependencies

• Python >= 3.10 if installed with pip or pipx
• To build debian packages with ops2deb build you need the following packages on your host:

sudo apt install build-essential fakeroot debhelper

If you plan to build packages for armhf and arm64 you will also need the following packages:

sudo apt install binutils-aarch64-linux-gnu binutils-arm-linux-gnueabihf

Getting started

In a test directory run:

curl https://raw.githubusercontent.com/upciti/ops2deb/main/ops2deb.yml
ops2deb lock # generate lockfile where downloaded file hashes are stored
ops2deb  # equivalent to ops2deb generate && ops2deb build

To check for new releases run:

ops2deb update

This command updates each blueprint in the ops2deb.yml configuration file with the latest version of the upstream application.

By default ops2deb caches downloaded content in /tmp/ops2deb_cache:

tree /tmp/ops2deb_cache

The cache can be flushed with:

ops2deb purge

For more information about existing subcommands and options run ops2deb --help.

Usage examples

Packaging kubectl

The fetch field tells ops2deb to download a file. ops2deb will check the hash of downloaded files against a lockfile. To generate/update this lockfile, run ops2dbe lock. By default, the lockfile is named ops2deb.lock.yml.

name: kubectl
version: 1.20.1
summary: command line client for controlling a Kubernetes cluster
description: |
  kubectl is a command line client for running commands against Kubernetes clusters.
fetch: https://storage.googleapis.com/kubernetes-release/release/v{{version}}/bin/linux/amd64/kubectl
install:
  - kubectl:/usr/bin/

Creating a metapackage

Ops2deb can be used to create metapackages:

name: allthethings
version: 0.1.9
architecture: all
summary: install various devops tools
description: Some great description.
depends:
  - kubectl
  - kustomize
  - helm
  - helmfile
  - devspace

Packaging ops2deb with ops2deb

Note that when the fetch key is not used, ops2deb will run the build script from the directory where it was called. Hence for the following blueprint to succeed, you have to run ops2deb from the root directory of this github project.

name: ops2deb
version: 0.15.0
homepage: https://github.com/upciti/ops2deb
summary: debian packaging tool for portable applications
description: |-
  Ops2deb is primarily designed to easily generate Debian packages for portable
  applications such as single binary applications and scripts. Packages are
  described using a simple configuration file format. Ops2deb can track new
  releases of upstream applications and automatically bump application versions
  in its configuration file.
script:
  - poetry install -E pyinstaller
  - poetry run task single_binary_application
  - install -m 755 build/x86_64-unknown-linux-gnu/release/install/ops2deb {{src}}/usr/bin/

Building packages for multiple architectures at once

If the upstream application is released for multiple architectures, use the matrix object to generate one source package for each architecture:

name: helm
matrix:
  architectures:
    - amd64
    - armhf
    - arm64
version: 3.7.2
homepage: https://helm.sh/
summary: Kubernetes package manager
description: |-
  Tool for managing Kubernetes charts.
  Charts are packages of pre-configured Kubernetes resources.
depends:
  - kubectl
fetch: https://get.helm.sh/helm-v{{version}}-linux-{{goarch}}.tar.gz
script:
  - mv linux-*/helm {{src}}/usr/bin/

The blueprint above will generate three packages: helm_3.7.2-1~ops2deb_armhf.deb, helm_3.7.2-1~ops2deb_arm64.deb and helm_3.7.2-1~ops2deb_amd64.deb

Note the use of the {{goarch}} variable which maps debian architectures to sensible go architectures.

You can also define your own architecture maps using the fetch.targets field and the {{target}} jinja variable:

name: bottom
matrix:
  architectures:
    - amd64
    - armhf
version: 0.6.6
revision: 2
homepage: https://clementtsang.github.io/bottom
summary: cross-platform graphical process/system monitor
description: |-
  A cross-platform graphical process/system monitor with a customizable interface
  and a multitude of features. Supports Linux, macOS, and Windows.
  Inspired by gtop, gotop, and htop.
fetch:
  url: https://github.com/ClementTsang/bottom/releases/download/{{version}}/bottom_{{target}}.tar.gz
  targets:
    amd64: x86_64-unknown-linux-gnu
    armhf: armv7-unknown-linux-gnueabihf
install:
  - btm:/usr/bin/

Using environment variables

You can use {{env("VARIABLE", "a_default")}} in all fields except fetch.targets.*. The example below uses environment variables set by Gitlab CI:

name: "{{env('CI_PROJECT_NAME')}}"
version: "{{env('CI_COMMIT_TAG', '0')}}"
homepage: "{{env('CI_PROJECT_URL')}}"
summary: awesome application for doing things
description: |-
  Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor
  incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis
  nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.
script:
  - install -m 755 build/x86_64-unknown-linux-gnu/release/install/great-app {{src}}/usr/bin/

Configuration file

Written in YAML and composed of a single blueprint object or a list of blueprints objects. A blueprint is defined by the following:

| Field | Meaning | Default | | -------------- | ---------------------------------------------------------------------------------------------------- | ------- | | name | Component name, e.g. kustomize. | | | matrix | Generate multiple packages from a single blueprint. | None | | version | Application release to package. | | | revision | Package revistion. | 1 | | epoch | Package epoch. | 0 | | architecture | Package architecture. | amd64 | | homepage | Upstream project homepage. | None | | summary | Package short description. | | | description | Package full description. | | | depends | List of package dependencies. Corresponds to Depends entry in debian/control. | [] | | recommends | List of package recommended dependencies. Corresponds to Recommends entry in debian/control. | [] | | conflicts | List of conflicting packages. Corresponds to Conflicts entry in debian/control. | [] | | fetch | A file to download. tar.gz, tar.xz, tar, zip and deb archives are extracted automatically. | None | | pre_script