Core Committer Weekly Interlock - May 30, 2017
Attendees:
Agenda
- 2.7.0 is red. Can we run master CI first in the process, check to make sure master is green before doing the sprint release? Former user (Deleted).?? For the most recent release of 2.7.0 plan is to get the release to green and release whenever that happens. Minor release number will change. We need a 2.7.1 for this week
- Amy Mullins (moved from 5/15 meeting) QRB discussed moving away from the internal EMC Code repo to a private github repo to address ease of use concerns (no longer required to require an account)
- This would move private lab configs and proprietary skupacks. Update - Internal code hub should be accessible to all Dell/EMC employees. AI Michael Hepfer can you check to see if Mustang can access?
- Amy Mullins Plans and process to ramp developers and expand merge capabilities across the teams. Update - Each CC per team will pair with folks on their team, once that CC person feels the team member is ready to gain merge access the CC member will grant the access. Please review RackHD Code Review and Merge Process by Friday.
- Former user (Deleted) RAC-5196: I would like to better understand the behavior for a workflow that fails but needs to do some clean-up
- Former user (Deleted) Moving PR gates to an external Jenkins instance. Should we bump this up in priority? Could help our latest issues on production Jenkins. - AI Former user (Deleted) Update where will we host it , or is our internal Jenkins sufficient but we are not utilizing it correctly?
- Former user (Deleted) To Vagrant Box or not to Vagrant Box? Do we know if anyone is using the Vagrant Boxes that we publish? It seems like more people are building their own for development. Update - Is there a reason why users can't go to Docker? (Anything that requires Vagrant?) AI Former user (Deleted) do you have metrics around # of downloads for the different deployment types? (i.e. is anyone downloading Vagrant?)
- Former user (Deleted) Are tests part of our release? Make sure we are all on the same page. Setting AffectsCommunity for test PRs. Update - Yes tests are part of the deliverable.
- Former user (Deleted) Timing of the tests in Jenkins. Should we decrease the frequency of ContinuousFunctionTest? Currently running every hour but competing for resources. Update ContinuousFunctionTest is running close to 1 hour for total runtime. For now leave it to every hour, recent changes now have only 1 running at a time should make this once an hour feasible. Need to keep an eye on it.
- Former user (Deleted) I see a lot more on-http Travis failures after the merge goes in but it runs successfully in the PR gates. Is this a known issue or one that needs to be investigated further.
- Former user (Deleted) Can we get more slaves for the Hopkinton Sandbox Jenkins instance? A lot of use in this Sandbox now (Yay!) but it is getting very VM constrained. Update - Former user (Deleted) can we get more slaves?
- Amy Mullins Deprecated hound checks have been removed last sprint (https://rackhd.atlassian.net/browse/RAC-5147).
- anything left to be considered before making hound checks a merge gate? AI James Turnquist need to get this aligned with Flake8. Plan going forward is to fix al Hound issues in PR's. AI All managers to communicate to their teams. AI Former user (Deleted), Former user (Deleted) Amy Mullins, Leo Zhang
- Amy Mullins Post package deployment testing
- If tests are passing FIT, need to clarify the expectations/guidelines for post package testing. Update AI Amy to bring up at QRB. (
- QRB discussion/suggestions for alleviating resource concerns on the production environment:
- Run MasterCI at a non working time. It was recently moved to 9am SH time. Suggest that it be moved to 6am to minimize resource allocation competing with the PR quality gates. Update - it is currently at 6:31 AM SH time, leave it as is.
- Turn off Docker CIT and OVA CIT in the post test. CIT is just about all converted with the exception of a few api test. Erika will be turning them on in the next 1-2 sprints. This will frees up 2 instances. - Update agree to plan
- Consolidate the 3 Docker OS Install to a single RackHD Slave instance where Post Package OS Install is executed on a single RackHD Slave instance in parallel. This will free up 2 slave instances.- Update agree to plan (Former user (Deleted) do you agree?)
- Consolidate the 3 OVA OS Install to a single RackHD Slave instance where Post Package OS Install is executed on a single RackHD Slave instance in parallel. This will free up 2 slave instances. - Update agree to plan (Former user (Deleted) do you agree?)
- This will drop the RackHD slave instances from 13 to 7.
- Former user (Deleted) it appears we have pipelines within pipelines is that required or can we go with single pipeline?
QRB question and AI for Former user (Deleted) to check with Former user (Deleted): What is the purpose of the Vagrant post test? Who is using the vagrant deployment and does it really need to be in the nightly build? What is currently running for this suite of tests.
- have we completed all Post testing stories?)