Closed gligneul closed 1 year ago
Just to get a better understanding of the context: the Inspect Server eventually recovers and can communicate with the Server Manager. The problem is just the cryptic error messages it sends while that doesn't happen. Is that correct?
Yes, precisely that. The inspect server does not maintain any state. So, once the server manager is available again, it will work.
Closing this one because we will rework the server-manager soon.
📚 Context
If the Server Manager is not available (e.g., it has been manually restarted), incoming requests will fail. At the moment, the error messages included in the response are very low-level and often meaningless to clients (e.g., "session id not found" or "'NoneType' object has no attribute 'process'")
✔️ Solution
This issue proposes that the Inspect Server should interpret the low-level errors returned by the Server Manager and produce more sensible responses to clients.
📈 Subtasks