Currently the REST-API doesn't log anything when an error occurs. I'm currently dealing with a problem where the lag between retries and me receiving the problem report is on the order of weeks. From the apache log (roundup runs as WSGI under apache in that application) I see that they receive a 400 from apache. Now I'd really like to know what they sent.

I propose to add logging to failed requests: Currently all errors are handled in the function _data_decorator in rest.py. So we could add a log there centrally and log the parameters of the request and the error message they receive.

I'm unsure about the severity of the log. Since this is the normal operation (and errors are not errors in roundup) I'd suggest to log this as INFO or lower. Or should this be a DEBUG log? I would run this always-on in production as it is normal operation that I have to debug a client web service on my end :-) So I tend to make this INFO.

What do you think?

Info sounds reasonable.