Cannot install kubeflow-metadata==0.3.1 because of ml-metadata

[ianhellstrom]

/kind bug

What steps did you take and what happened: I cannot install kubeflow-metadata version 0.3.1 because it requires ml-metadata version 0.21.1, which does not appear to be available through pip.

pip install kubeflow-metadata==0.3.1
Collecting kubeflow-metadata==0.3.1
  Using cached kubeflow-metadata-0.3.1.tar.gz (10 kB)
ERROR: Could not find a version that satisfies the requirement ml-metadata==0.21.1 (from kubeflow-metadata==0.3.1) (from versions: 0.12.0.dev0, 0.13.0.dev0, 0.13.1.dev0)
ERROR: No matching distribution found for ml-metadata==0.21.1 (from kubeflow-metadata==0.3.1)

pip install ml-metadata==0.21.1
ERROR: Could not find a version that satisfies the requirement ml-metadata==0.21.1 (from versions: 0.12.0.dev0, 0.13.0.dev0, 0.13.1.dev0)
ERROR: No matching distribution found for ml-metadata==0.21.1

What did you expect to happen: Install kubeflow-metadata as requested.

Environment:

• Metadata version: 0.21.1
• Kubeflow version: 1.0.1
• Kubernetes version: 1.18.6
• OS: MacOS 10.15.6
• Python 3.8

[issue-label-bot[bot]]

Issue Label Bot is not confident enough to auto-label this issue. See dashboard for more details.

[ianhellstrom]

This appears to be related to: https://github.com/google/ml-metadata/issues/37

[davidspek]

I'm also running into this issue while trying to add multi-user env support to kubeflow-kale. The Notebook Server I am running is using python 3.8, which is supported in ml-metadata 0.24.0 released a week ago. https://github.com/google/ml-metadata/releases The only breaking change that is listed when updating from 0.21.0 to 0.24.0 is Python MetadataStore APIs return mlmd errors instead of tensorflow errors.. Would it be possible to simply update kubeflow-metadata to use ml-metadata 0.24.0?

EDIT: I just installed the python SDK using ml-metadata==0.24.0 and I could run through the demo notebook without any issues.

[issue-label-bot[bot]]

Issue-Label Bot is automatically applying the labels:

Label	Probability
area/jupyter	0.68

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