Agenda

Sprint Review

  • Review the Sprint objectives
  • Review what was accomplished
  • If applicable, demo the work that was done
  • Discuss what needs to be done
    • Feedback from Jon Crabtree
      • Currently unable to verify or recreate reported issue(s)
      • Environment is unstable 
      • We need to test more 
      • Worked down to the last minute and didn't allow sufficient time for testing
      • We should try a latency test
    • Does this work qualify as “developer support” or is this a separate thing?
Task/FeatureStatus
  • Bugs - refresh
Done
  • Ingress/load balancer
 
    • Prototype install/evaluation
Done
    • Put into a reproducible method (deploy)
  • Manual method is done
  • Ansicble task isn't done
    • GUI/API server changes
Done
    • TLS
  • Done
  • However for a production certificate, will need to complete security review
  • Storage:
 
    • put into a reproducible method
  • Done
    • Kuberenetes integration
  • 95% complete, bugs were discovered that need to be addressed
  • Multinode cluster: 
 
    • Need to make sure the contrib/ansible playbooks do what we need
Done
    • Automate OpenStack setup
Done
    • 1 master, 3 etcd, nodes as needed, 4 storage
Done
    • Storage integration in Kubernetes

Done

  • Console into running containers
Done
    • wss
Done
  • Memory limits in specs
Done
  • Testing 20 users
Done
  • Traces/profiling during Odum
Done



Sprint Retrospective

  • What went well?
    • Managed to complete sprint despite people out of the office or occupied with other projects
    • Routine/process is coming together
    • Risk manifest itself, but still got a lot of stuff done
    • Encountered and overcame non-trivial hurdles
    • Nice balance of skills in the team
    • Developed working relationship with the Nebula team
  • What went wrong?
    • Over-committed to the point of frustration
      • Knew there were risks, had a plan b, wanted to push to get things done but couldn't because of people out of the office
      • Shutting down on the 20th and going with plan B, instead of trying to push through until the last moment possible
    • Technology is new and moving - figure out if we need the newest and greatest
    • Nebula 
      • a log file filled up that slowed things down
      • Couldn't run tests in parallel for a week
    • We have a definition and estimation problem
      • Estimation is hard because the definition isn't well defined
      • Account for unforeseen issues
    • Tasks sat in review
    • Workflow not universally understood and accepted
  • What could we do to improve?
    • Better job of defining the tasks
      • Each person defines their own tasks
    • Better job of estimating the tasks
    • Don't commit unless you know exactly what needs to be done
    • Commit must include review tasks
      • Part of your job is to review other tickets
    • Pair reviewing when possible 
    • Don't pull an item into Sprint until all developers understand the work that needs to be done
      • This will help with handing off tickets (or taking on additional work)
      • This will help with reviews
  • Commitments:
    • Keep doing
      • Keep following and refining the process (everything being done)
    • Start doing
      • Better planning
        • Account for review gap
        • Account for planning gap 
        • Each person defines their own tasks (provide time in the sprint for this) 
      • Better estimating
    • Stop doing
      • Overcommitting 
    • Everyone review the workflows, meet and refine as needed
  • No labels