See #1 and #2.
For analysis and bug hunts, we need the messages to know if it's something our side or to just catch trends in external server implementations to update the link checker.
I would note also that we need to handle exceptions related to the request (encoded as 900s in the initial analysis) for the same - so not an HTTP status response but a code exception that relates to some issue with the requests (timeouts, some SSL cert problem, etc).
See #1 and #2.
For analysis and bug hunts, we need the messages to know if it's something our side or to just catch trends in external server implementations to update the link checker.
I would note also that we need to handle exceptions related to the request (encoded as 900s in the initial analysis) for the same - so not an HTTP status response but a code exception that relates to some issue with the requests (timeouts, some SSL cert problem, etc).