March 2019 Endgame

[scarlett2018]

This is the endgame schedule for March 2019 Iteration.
last updated on 3/21/2019

Schedule

• End game master: @mzmssg
• March 15 20, 2019 Code freeze for the endgame
• March 29 April 3, 2019 Endgame done
• March 29 April 3, 2019 Expected release date (this may change)

Friday 03/15 Wednesday 3/20 @mzmssg

• Code freeze at 6pm UTC+8
• Release branch pai-0.11.y

Monday 03/18 Thursday 3/21

• Image build @mzmssg
• Int bed deploy @mzmssg
• Feature test case ready
  • Detect and alert for unhealthy GPU @mzmssg
  • Provide the ability to query all the jobs in a Node in PAI Web Portal @xudifsd
  • Team wise storage management support @wangdian
  • Simplest Job Submission Page for Job (types) that will only run on 1 node @debuggy
  • Provide filters for jobs list page @Gerhut
  • Refine Job detail page @sunqinzheng
  • Basic supports for reserving container's life time for debug after job completion @ydye
• Test @mzmssg

Tuesday - Wednesday 03/19 - 03/20 Friday - Monday 3/22 - 3/25

• Release checklist validate @mzmssg
• New feature test (test owner to be decided)
• Bug bash
• Cherry-pick hotfix
• Remind teams to verify issue status

Thursday 03/21 Tuesday 3/26

• Release note ready, confirm with team members. @scarlett2018 @mzmssg
  • Each release should update the quick-start template to using the latest release image tag.
  • Each release may need to update the upgrade tutorial, for example, change version numbers in some place or so on.
• Rebuild images @mzmssg
• Push images to public repo @mzmssg
• Prod bed upgrade headsup announcement @scarlett2018

Friday 03/22 Wednesday 3/27

• Prod bed data backup @mzmssg
• Prod bed upgrade @mzmssg
• Polish release notes @scarlett2018

Monday-Wednesday 03/25-03/27 Thursday - Monday 3/28 - 4/1

• Dogfood user test @scarlett2018
• Collect feedback @scarlett2018
• Fix blocking issue
• Polish release notes @scarlett2018

Thursday-Friday 03/28-03/29 Tuesday - Wednesday 4/2 - 4/3

• Add git tag @mzmssg
• Release announcement @scarlett2018

[hao1939]

Add two items(May go along with the release-note PR):

1. Each release should update the quick-start template to using the latest release image tag.
2. Each release may need to update the upgrade tutorial, for example, change version numbers in some place or so on.

[mzmssg]

Should change the version to the latest and adapt this script
https://github.com/Microsoft/pai/blob/86f0fd50fa2c0f47d359824d8e80bb9a93ef0665/deployment/tools/configMigration.py#L66