Container SIG/Flock2018-notes

Notes from Flock 2018 container workshop

• How do you get the new build to bodhi?
  • Code is done.
  • Can push images to registry
  • Needs to pass final testing.
• Multiarch images are coming soon
  • intel, arm, power, s390
• Betka (sync upstream content to downstream)
  • fedmsg -- parse messages
  • pagure webhooks
• Flatpak builds container images from module dist-git.
• Build container images automatically out of modules.
• Pingou wrote koji-simple-ci tool to build RPMs as scratch builds in koji.
  • Get inspired for the same thing for container images.
• 2 weeks rebuild: we should have an automation in-place to create bodhi updates.
  • A service which would create bodhi update once a container image build is done (via fedmsg).
• TODO: open source the bots finally!
  • Create tasks in container SIG and get help from community.
• Enable using candidate registry as a base image for builds.
• SIG
  • We need a leader who would organize the whole SIG -- connect people, coordinate, decide things.
  • Focus
    • Guidelines
      • Review existing guidelines (an easy way to submit and maintain container images)
    • https://github.com/projectatomic/ContainerApplicationGenericLabels figure out status
    • https://github.com/container-images update template repo and status of the images
    • Inftrastructure
      • Run on top of Fedora based one (using packages and images build on our infrstructure).
      • tools, etc
    • Support the container users
      • image streams,...
    • Define the tasks well so anyone can pick them up.
  • issues
    • not @packager, but @container-maintainer
    • time and place of the IRC meeting
      • #fedora-containers
    • write down best practices for container SIG: where do we discuss what?
      • Fedora CoreOS: discourse - users, pagure - design discussions
    • Fedora taiga?
    • Issue tracking for container images
      • Dusty suggests pagure dist-git issue tracker
      • Clement uses WG issue tracker
• F30 is the time when atomic thing will start going away
• CentOS has a container something SIG (CCCP) -- do we join forces?
  • Invite them for our SIG meeting
• Validation.
  • How can I use taskotron to validate my image?
  • Dockerfile linting
  • Does it run in OpenShift? (Do we care about this?)
• Automation (least amount of steps for container maintainers).
• We could get people to submit PRs to update the Dockerfiles in distgit since I think there are a few that are a bit out of date.
• Automatic versioning based on a "main rpm" version (maybe we work on that during hacking time)