[Bug]: event log empty for new notebook that is failing to start

[christianvogt]

Is there an existing issue for this?

• I have searched the existing issues

Deploy type

OpenDataHub core version (eg. v1.6.0)

Version

1.7.0

Current Behavior

Maybe some relation to #579

My notebook server is not starting and the UI did not display any error message or event logs as to why. Dropping back to the console was the only way to understand that the dev cluster had a resource issue.

Expected Behavior

Event log and / or pod status should be presented to the user to understand why it's not starting.

Steps To Reproduce

Using a new dev cluster which has limited resources apparently.

1. create a new project
2. create a new workbench and use a large size. Consider adding a new notebookSizes entry to the OdhDashboardConfig if your notebook server still starts.
3. Observe the pod not starting.
4. Check the dashboard UI by clicking on Starting... and viewing the event log

Workaround (if any)

none in the dashboard

What browsers are you seeing the problem on?

Chrome

Anything else

No response

[christianvogt]

The Notebook instance has no notebooks.kubeflow.org/last-activity annotation and therefore the UI assumes it hasn't started.

Here's what the UI should show:

[lucferbux]

I think #793 is also related to this one right?

[christianvogt]

I think #793 is also related to this one right?

Certainly the pod failing as described in that issue is similar with the same outcome. However that issue does talk about the expectation to look at more than just the pod for relevant status. So at least in part they should share the same resolution for the immediate use case described.

[christianvogt]

@lucferbux unfortunately #793 will require additional changes because the error occurs on the StatefulSet before a pod is created. I tested a fix for this issue and it had no affect on #793