QA (public) » openQA Project (public) » openQA Infrastructure (public)

Motivation¶

Every six months we update every single machine. Updates are performed based on steps to copy out of our wiki. There is ambiuity and undocumented assumptions and potential for error e.g. not looking at or saving .rmpnew files.

Acceptance Criteria¶

• AC1: Scriptability of our documented steps has been researched
• AC2: Gaps in our documentation are known and filled in

Suggestions¶

• Implement a script that follows the steps to upgrade the system
  • Collect all logs for review in case there's issues after the upgrade
  • Cross-check how we/others review changes
  • Document gaps in the flow
  • If/ how to re-trigger deployments or how running openQA jobs are handled is not documented
  • How much do we care abuot .rpmnew files
  • Do we care that running tests break? Maybe not
• Semi-related idea
  • Send an automated email with artifacts from the update/ version details
• Implement a workflow in salt/ deployment pipelines or somewhere else
  • ansible would be better as we get a chance to see that this is supported (for the benefit of testing non-salt deployments)
• Mention to stop worker slots before doing an update in our documentation