Add logging so when bad things happen, we have some sort of record (if messages are collected from the workers!).
While this fix was going on the UChicago S3 interface was upgraded - so even without the retry it worked. In timing measurements this seemed to make no difference, however, so it is left in for when we go back to the older configuration.
fsspec
infrastructureretry
callbackWhile this fix was going on the UChicago S3 interface was upgraded - so even without the
retry
it worked. In timing measurements this seemed to make no difference, however, so it is left in for when we go back to the older configuration.Fixes #44