roadmap.md

TF Operator Roadmap

0.1

• Lead: @______
• Planned release: March 30, 2018

Phase 1 (Now - March 15, 2018)

• Refactor the code and job status
  • Refactor TFJobStatus in CRD API kubeflow/training-operator#333
  • Make the TfJob controller more event driven kubeflow/training-operator#314
  • Deprecate the TfImage field kubeflow/training-operator#330
  • Deprecate TfPort and set default port for users kubeflow/training-operator#327
  • Manage Pods directly instead of using Job controllers kubeflow/training-operator#325
  • Don't leave pods running just to get logs kubeflow/training-operator#128

Phase 2 (March 16, 2018 - March 30, 2018)

• Improve the test cases for local TF jobs
  • [enhancement] Add e2e test cases for recorder kubeflow/training-operator#317
  • Increase the unit test coverage (27% -> 50% or above)
  • Add more e2e test cases for exception paths

Low-priority Features

• Use headless services for Training jobs kubeflow/training-operator#40
• Set a default value for restartPolicy kubeflow/training-operator#55
• Structured Logging For the operator kubeflow/training-operator#24
• Use conditions instead of phase kubeflow/training-operator#223

草泥马 我肯定会用行动让天朝封了这个社区，滚回大中国局域网写你的红色代码吧