Description of problem:

In some cases where the spawner is not able to start a server pod for the user, the spawn modal can get stuck indefinitely without showing any error message or failure notification.

This has happened in a long running automated test, and it appears that the modal was stuck with the loading bar empty for multiple hours.

A good way to reproduce this behaviour seems to be trying to spawn a server with a size bigger than what is available on the cluster (e.g. by default Large/XLarge should trigger this)

The user can always click on "Cancel" to close the modal and try spawning again. However I think that it's not good practice to not communicate anything to the user (ever) and not inform them of why the server is not starting.

In JH we used to show the error message from OCP of there not being nodes available with enough requested cpu/ram/etc., and while it wasn't pretty it at least gave a pretty good idea of what the problem was.

Prerequisites (if any, like setup, operators/versions):

RHODS 1.16 RC on OSD

Steps to Reproduce

1. Install 1.16 RC
2. Go to spawner page
3. Set size to something bigger than what is available (on a default cluster both Large and XLarge should work)
4. Click on "Start server"

Actual results:

The modal is indefinitely stuck with no progress on the loading bar, and does not show any error message in the default view or by expanding the event log

Expected results:

There should be a hard timeout after which the modal informs the user that is was not possible to provision their requested server

Reproducibility (Always/Intermittent/Only Once):

Always

Build Details:

RHODS 1.16 RC

Workaround:

The user can always click on "Cancel" to close the modal and try spawning again.

Additional info: