It added a lot of noise to the logs. If there's a problem with waiting
for requests it will still raise an error when it times out so we
should still be able to trace back to the cause of the problem.