Hmm, there is a check in the backend for editing a closed request.

My hypothesis would be that the PUT request to cancel the request happens twice rather than once. The request being cancelled suggested the first one goes through, the second fails.

Verifying that would likely require either the logs or using the front developer tools to check if multiple HTTP requests have been made

and I know it's difficult to do much with bugs that don't repro consistently, but I thought you might have a theory as to what is happening since you both have been looking at this code pretty closely lately.

Is there anything we should ask Chalmers (or others) to do or take note of when they see the issue? I have seen this issue myself many times but not more than once every few days and I can never repro after.