Puppet OpenStack success during Mitaka cycle

The winter is almost finished and Mitaka cycle is about to end. It’s time to celebrate our success and get prepared to our next challenges.


The team is scaling-up

One of the plans for Mitaka was to scale-up the core reviewer team; we wanted to give a chance to anyone involved in the group to get rewarded by becoming core-reviewer.
During the cycle, we were very happy to have more core-reviewers :

  • Denis, Cody and Alex became core on all modules.
  • Sofer became core on puppet-keystone.
  • Sergey became core on puppet-neutron.

Creating teams for some modules is a success for us, we give the opportunity to people involved on a specific OpenStack project and its Puppet module to become core and get involved quickly, it’s something we might want to continue. It helps to establish collaboration and trust between different groups, with the same goal to improve OpenStack all together.


Our CI is “up to eleven”

Collaboration works, and our project is a good example.

  • OpenStack Infra
    • CI jobs optimization to save resources
    • Enabling voting for integration jobs
    • Collaboration between OpenStack Infra & Puppet OpenStack groups for Puppet jobs
  • Tempest
    • Puppet Jobs are now in Tempest Experimental queue and can be run on demand
    • Keep reporting bugs and giving feedback
    • Integration with openstack-health
  • Packaging
    • Providing early feedback to RDO community by testing trunk packages
    • Helping Canonical to test Ubuntu packaging for Mitaka
  • OpenStack Installers
    • Run TripleO jobs for Puppet OpenStack modules CI
    • Moving forward with Fuel
  • More testing coverage
    • more projects tested thanks to Tempest plugins
    • more use-cases to cover the different backends / drivers but also SSL & IPv6 deployments

This collaboration is essential for a project such as Puppet OpenStack, we need to keep it and continue the work here. Our CI coverage is I think one of the reasons why our modules are stable and adopted by a large number of people in OpenStack community.


Getting started becomes easier

One thing we mentioned in Tokyo was the need to have a tool to quickly deploy an all-in-one OpenStack setup with our Puppet modules. We made it with a simple shell script, where anyone can deploy OpenStack in a few minutes with our modules. Our documentation also got improved, so we can welcome newcomers more quickly.


Our diversity keeps growing

Looking at Liberty & Mitaka stats, contributions metrics (patches, reviews, bugs) did not change much, though we keep having more diversity and welcome new contributors.
Our contributors are usually working as OpenStack operators or / and developers:

  • OpenStack Installers: TripleO, Fuel, Packstack
  • OpenStack Operators: Time Warner Cable, Internap, GoDaddy, Puppetlabs, OpenStack Infra Cloud

Having people from both worlds keep reducing the feedback loop and helping to build strong Puppet modules, that work on production deployments at scale.


More challenges

I often remember this thing:

La route est longue mais la voie est libre.

French statement from Framasoft, which could mean:

The road is long but the path is clear.

Together, we have been through different challenges. Some of them were about collaboration, others were technical.
The future looks great to us, here is an overview of what I think is our next challenges:

  • With the huge number of Puppet modules (~30), we need to keep them updated, tested and documented. More modules will come in Newton.
  • Investigate Multi-Node Orchestration.
  • Switch functional testing to newer versions of Puppet.
  • Keep growning our team but also the collaboration with external projects.

These challenges will help us continue to be a reference in OpenStack deployment tools and I’m looking forward to celebrating it again, with you, in 6 months.

Software Engineeer at Red Hat, Private Pilot, French guy hiding somewhere in Canada.