When admin switches the engine to DevWorskpace all che-server based workspaces should be stopped

[ibuziuk]

Describe the bug

• install 'eclipse-che' using the stable channel
• create CR with a single-host and OpenShift Auth enabled
• create an account (using 'developer' OpenShift account on the crc for example) with a couple of workspaces and start one of them
• wait until workspace is running, run build / run command
• update the CR from the eclipse-che namespace and enable devWorkspace

  devWorkspace:
    enable: true

• go back to the running workspace - UD does not show it anymore, however workspace pod is still running
• ERROR: previously running workspace pod is still running, even though UD does not show it

Che version

7.37@latest

Steps to reproduce

• install 'eclipse-che' using the stable channel
• create CR with a single-host and OpenShift Auth enabled
• create an account (using 'developer' OpenShift account on the crc for example) with a couple of workspaces and start one of them
• wait until workspace is running, run build / run command
• update the CR from the eclipse-che namespace and enable devWorkspace

  devWorkspace:
    enable: true

• go back to the running workspace - UD does not show it anymore, however workspace pod is still running
• ERROR: previously running workspace pod is still running, even though UD does not show it

Expected behavior

When admin switches the engine to DevWorskpace all che-server based workspaces should be stopped

Runtime

OpenShift

Screenshots

When admin switches the engine to DevWorskpace all che-server based workspaces should be stopped

Installation method

OperatorHub

Environment

Linux

Eclipse Che Logs

No response

Additional context

No response

[l0rd]

But the way to switch to the DW should be to use the all-namespace channel rather than updating the CheCluster CR (although this issue will still be there, but switching to all-namespace channel is important to avoid other kind of problems).

[ibuziuk]

This issue happens after the switch from stable to all-namespace - more details in #20614 (comment)

[ibuziuk]

PR has been merged eclipse-che/che-server#177
Closing

[nickboldt]

So with this change, once a CRW instance is updated from 'latest' (no DW) to 'all-namespaces' (DW enabled)...

• existing workspaces are stopped

How does a user restart the workspace? Will it be automatically converted to v2.1 for them?

Or, will they have to manually fix things, in one of two ways:

a) manually update the devfile in the dashboard editor to upgrade it to v2.1 (and then commit these changes back to their source repo so changes are persisted)

b) open a NEW workspace from the existing devfile via the factory URL, and have it converted for the user automatically... then user needs to remember to commit the updated devfile back to the source repo so the converted version 2.1 flavour instead of the now-obsolete 1.0 flavour is kept and can be reused.

Are those the correct user stories?

[dmytro-ndp]

@nickboldt : there is an issue about further actions: #20845