[Release 1.1] training operator release

kubeflow / common

Common APIs and libraries shared by other Kubeflow operator repositories.

Apache License 2.0

51 stars 73 forks source link

[Release 1.1] training operator release #97

Closed Jeffwan closed 3 years ago

Jeffwan commented 4 years ago

Opening this issue to track releasing training operators for Kubeflow 1.1 and would like to add area OWNERS for @johnugeorge @terrytangyuan @Jeffwan

Seems we want to these works done before 1.1 release. We may not finish all of them but let's have a try.

Migration operator implementation to kubeflow/common fashion. (TF, PyTorch in pending)
- https://github.com/kubeflow/tf-operator/pull/1171 @ChanYiLin
- https://github.com/kubeflow/pytorch-operator/issues/282
Graduate Operator to v1
- MXNet Operator https://github.com/kubeflow/mxnet-operator/issues/66
- XGBoost Operator https://github.com/kubeflow/xgboost-operator/issues/44
- MPI Operator https://github.com/kubeflow/mpi-operator/issues/217

issue-label-bot[bot] commented 4 years ago

Issue-Label Bot is automatically applying the labels:

Label	Probability
area/operator	0.86
kind/feature	0.79

Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback! Links: app homepage, dashboard and code for this bot.

kf-label-bot-dev[bot] commented 4 years ago

Issue-Label Bot is automatically applying the labels:

Label	Probability
area/operator	0.86
feature	0.79

Please mark this comment with :thumbsup: or :thumbsdown: to give our bot feedback! Links: app homepage, dashboard and code for this bot.

terrytangyuan commented 4 years ago

@Jeffwan Shouldn't we go through some sort of review process before the graduation of operators though? I don't think we would have sufficient time to finish that before 1.1 release.

jlewi commented 4 years ago

the apps cd pipelines have been fixed
application owners please review PRs to update your apps and ensure they get merged
- https://github.com/kubeflow/manifests/pulls/kubeflow-bot

/cc @terrytangyuan @Jeffwan

Jeffwan commented 4 years ago

@terrytangyuan

Agree. I think once these operator meet done criterion, then they can be graduated. It would be better for us to connect owners and have some discussion. If we can not catch 1.1, I think it's fine.

jlewi commented 4 years ago

@terrytangyuan @Jeffwan any update?

terrytangyuan commented 4 years ago

I think we should be ready regarding manifests (@Jeffwan correct me if I am wrong). Although There seems to be more pending auto PRs. @Jeffwan Could you check whether they need to be merged to master and cherry-picked to 1.1 release branch?

Jeffwan commented 3 years ago

1.1 and 1.2 release are done. We may not use exact same version as Kubeflow in the future. WG operators will have their own branch or tags. We just need to make sure we bring the right version to kubeflow release.