If SSE fails and we're in the sseFailure state, should we be taken out of ready? Currently it looks like that will happen but I'm guessing we can still show the timeline (and even send build requests though that may be unwise). I wonder if you want to have a workspaceAndResources submachine inside ready with listenForEvents and sseFailure as its states.

How should we handle SSE errors? Should the page show something about failing to refresh data? Maybe a warning-style thing rather than error-style, like when your workspace is going to stop soon?

How should we handle SSE errors? Should the page show something about failing to refresh data? Maybe a warning-style thing rather than error-style, like when your workspace is going to stop soon?

The backend will close the connection when we get a non-recoverable error. My first thought is to console log the errors we get or something low-stakes like that, and just keep it open unless the backend closes it. A non-fatal error may be a momentary db issue, but the next message may come through fine. On the connection being closed we should retry, probably with a backoff.

If SSE fails and we're in the sseFailure state, should we be taken out of ready?

@presleyp good catch; this should be fixed.

How should we handle SSE errors? Should the page show something about failing to refresh data? Maybe a warning-style thing rather than error-style, like when your workspace is going to stop soon?

@presleyp Right now, SSE errors will not block the user from viewing the page. I will put up a subsequent PR with a new, warning style message indicating workspace data might not be up to date if that field is populated. I want to handle that separately because there are several workspace error cleanup items to tackle and it would be nice to have them all in 1 PR.

On the connection being closed we should retry, probably with a backoff.

@f0ssel Yeah, we should add a backoff. Right now, we close, wait a second and then try again with a new connection. I would like to add that backoff when I add the warning described above.