Historically compass can handle connections being intermittently unavailable, so all the logic around connection scoped workspaces is based on the idea that the connection is still alive, even if it sometimes can error out.

With compass-web we're starting to run into specific cases where existing connection can just completely disappear with no way of recovering and we want to somehow specifically account for that state so that the tabs are still on the screen, but we don't allow user to do anything that would lead to excessive errors and would want to indicate that the tab is disconnected.

This ticket is about investigating a bit more what's the expected outcome (maybe the current state where actions just error out is fine) and the effort that would be needed for implementing this.

TRIAGE UPDATE: This ticket is not meant to handle the case where a user has been disconnected due to an Atlas user role change or their Atlas auth session expiring. This is only applicable to when the state of the cluster changes, and they're not able to connect anymore (i.e. cluster paused, deleted, restarting). We won't be able to make a tab active/accessible again unless we do this ticket without the user refreshing (same applies if a role change takes away access & then regrants it)