Public Cloud Robot Suite

Proposers

• Gerardo Garcia (Telefonica)
• Mark Beierl (Canonical)
• David Aguilar (Indra)

Description

Currently the daily jobs have a specifically designed workflow that:

• Builds docker images
• Pushes them to private GitLab docker registry
• Installs OSM from the private GitLab registry
• Starts a robot container to:
  • Create the VIM for testing
  • Create the K8s cluster for testing
  • Create the SDNC controller for testing
  • Run a suite of tests
• On success, push all images to dockerhub

What we want is to examine and design a new structure to decouple

• Deployment of OSM, so the same OSM can be reused for different robot runs (ie: Openstack, then Azure)
• Creation of the VIMs and K8s clusters
• Running of a suite based on a vim target name (so we can choose between different clouds)
• Publication / merging of robot results
• Publication/push to dockerhub on aggregate result success (all clouds should pass before we publish the latest image)

Demo or definition of done

Our daily job (or a new weekly job) should run using the new pipeline structure. The job should run tests in ETSI Openstack VIM and in Azure.