Python/Flask integration stack trace is not ALWAYS included with logger.exception

@erik-farmer yeah. In that case I’m not surprised this happens. I believe it happens if a context switch happens upon error handling. I have seen this before and the best workaround I have is to do this:

try:
    ...
except Exception:
    exc_info = sys.exc_info()
    logger.error('message here', exc_info=exc_info)

It still does not solve it in all situations but lowers the risk of the traceback going away.