Page tree
Skip to end of metadata
Go to start of metadata

Date

Attendees

Discussion Items

Euphrates Release 1.0

  • Branch has been created
  • Daily jobs and docker builds have been created (status page: Build Status)
  • Looking into getting a daily under ARM with FUEL

F Release Kickoff

  • Scale out of storage testing:
    • One idea is to support multiple stacks with multiple tests
    • Another idea: create stack of N VMs (say 20), and then run test using only X of them (say 10).  Then re-run with more or less
    • Creating multiple stacks would provide better isolation, specifically with the nova availability zone support, agreed that we should pursue this option.
    • Take a look at how Yardstick uses Heat and move towards an object approach to stack creation vs. direct upload of heat template and variable substitution
    • Two options:
      • Clone the code and modify it to our use (danger is not keeping up, requiring maintenance if bugs are found, new features are added)
      • pip install functest and yardstick in our environment (ie: dev env, and docker build).  Danger is if Yardstick or Functest makes a change, we could break
    • Benefit of having Functest code is we can eliminate the code in StorPerf to push results to DB.
    • See functest dockerfile:

          wget -q -O- https://git.openstack.org/cgit/openstack/requirements/plain/upper-constraints.txt?h=$OPENSTACK_TAG | \
              sed -E s/^tempest==+\(.*\)$/-e\ git+https:\\/\\/github.com\\/openstack\\/tempest@\\1#egg=tempest/ \
              > upper-constraints.txt && \
          pip install --no-cache-dir --src /src -cupper-constraints.txt \
              -chttps://git.opnfv.org/functest/plain/upper-constraints.txt?h=$BRANCH \
              git+https://gerrit.opnfv.org/gerrit/functest@$BRANCH#egg=functest && \
      
      
    • We would need to put a hard-coded value for BRANCH in order to protect against changes, until we are ready for the next version.
    • For development, use BRANCH=master unless this is changing too rapidly
    • Poll: how do we want to organize who is working on what?
      • Self- assign, or
      • Pre-assign to each person?
    • Decision is to leave everything unassigned and team members will assign as they are ready to work on something and then others know what is available, and who is working on what.
    • Mark Beierl to create JIRA stories for items in F release.
    • Team members are invited to create their own JIRA items as needed or as ideas come up.
  • Barometer co-ordination:
    • Mark Beierl what does this mean?  Write up proposal on how to use the two together and what we would like to see.  Others are free to take this and move it forward.

 

  • Mark Beierl  to set up session for walkthrough of the StorPerf code and how the current stack is tied to the single storperf-master
  • Mark Beierl open JIRA story for Functest/Yardstick pip install
  • Taseer Ahmed to give quick (15m) overview of XCI and how we can fit into it.  From there we can do a breakdown of tasks that are needed for StorPerf to fit into that pipeline.
  • Mark Beierl open JIRA for integration of reporting with TestResults DB dashboard.