Announcing Vagga 0.8.1

We are excited to announce the next version of vagga, the containerization tool that makes setting up development environments easy and fun.

This release brings local overrides for vagga commands, support of ubuntu bionic, better CI support, space reuse between multiple projects, and more.

Also while we were preparing this release vagga have got a logo:

Space Optimizations

Because containers contain a whole system, they are often quite large. Some of our users have over than a hundred projects with multiple containers in each of them. Previously, we allowed hardlinking files from multiple container versions in the same project.

Since vagga 0.8.1 the following config can save you more space:

storage-dir: ~/.vagga/images
index-all-images: true
hard-link-identical-files: true
hard-link-between-projects: true

We don’t enable it by default because malicious containers can infect files of other containers. But we explore other techniques to provide same space usage and also security (for example using reflinks instead of hard links) so stay tuned!

Local overrides

You can now add vagga.local.yaml or .vagga/local.yaml file to the repository and put containers and command into it. It works the same as any other mixin, but it has priority (i.e. overrides commands defined in normal vagga.yaml and its mixins)

This can serve several purposes:

  1. Add a vagga.yaml to the project that doesn’t have one (i.e. put the file .vagga/local.yaml, and put .vagga in global git ignores)
  2. Have your own commands that don’t make sense to everyone else, for example vagga publish if you are the only person who can publish. Or add an extra external volume having a new version of a dependency.
  3. Generate vagga.yaml itself, including bootstrapping the project.

As an example of bootstrapping here is very simplistic bootstrap script:

containers:
_bootstrap-alpine:
setup:
- !Alpine v3.8
commands:
_init: !Command
container: _bootstrap-alpine
run: |
echo -n "Project name: "
read name
cat <<END > vagga.yaml
containers:
alpine:
setup:
- !Alpine v3.8
commands:
run: !Command
description: "Run $name"
container: alpine
run: [echo, "Running $name"]
END

Then put file online and use it for seeding vagga.local.yaml:

$ curl http://example.com/bootstrap.yaml > vagga.local.yaml
$ vagga _init
Project name: hello
$ vagga
Available commands:
run Run hello

For a real project, it’s expected to use cookiecutter or yeoman or any other tool for bootstrapping. The vagga.local.yaml contains ordinary containers so you can use any scaffolding tools seamlessly without any extra setup on a host system. Bootstrapping even works when using vagga-box.

CI Support

We have a few new settings that are expected to be used on continuous integration systems:

propagate-environ allows to propagate some environment variables to the vagga commands automatically. For example, the following works good for Gitlab CI, so that usual variables like CI_JOB_ID and CI_COMMIT_SHA are available in commands:

propagate-environ:
- CI
- "CI_*"

ubuntu-skip-locking: true allows building images in parallel even when using same ubuntu package cache

versioned-build-dir: true allows building multiple versions of the same container simultaneously (when run with the same storage dir)

Note: these are settings. I.e. they are put into $HOME/.config/vagga/settings.yaml on a CI system itself rather than in vagga.yaml

Smaller Improvements

  • group option allows to commands in the help grouped by a title
  • vagga --version now works without vagga.yaml
  • Added rules param to !Build (similar to !Copy, see previous release notes for examples)
  • Added vagga --containers-only some_command which is useful to prepare all the containers needed to run some command(s). This is often useful for building tools on top of vagga

You can find more in release notes and upgrading guide.

Contributors

Release 0.8.1 consists of 160 commits from 4 contributors. Thank you!

  • Alexander Koval
  • Alex Kuzmenko
  • Paul Colomiets
  • Rainboh