We upload an almost 50G file with a python script to the fedora repository. And we had to abruptly cancel it (Ctrl + C), since we needed a downtime on the repository server.
Now we can not upload the file again. I do not see the file specified in the metadata of the xml object. When loading, I get the following error:
Traceback (last last call):
File "create_datastream2.py", line 386, in
dsresponse = s.send(prepped)
File "/opt/python_2_7_11/lib/python2.7/site-packages/requests/sessions.py", line 576, send
r = adapter.send(request, **kwargs)
File "/opt/python_2_7_11/lib/python2.7/site-packages/requests/adapters.py", line 426, send
raise ConnectionError(err, request=request)
request.exceptions.ConnectionError: ('Connection aborted.', error (104, 'Connection reset by peer'))
And I can upload any other file without any problems. It is in this place where we suddenly stopped that he fails.
Any ideas? I can share the script download with you if you need to.
Update 1: Actually, now I remember, I deleted this file from the web portal, thinking that it could be a partial file. And this created a tombstone and prevented the reloading of another file with the same name.
To fix this, I only had to remove this headstone using the DELETE query.
So, I think this does not cause any problems if the file upload is abruptly canceled? Any ideas what happens to the partially downloaded file ?. Does it have any leftovers and takes up some disk space? And java.io.tmpdir points to a disk with enough disk space (1 TB). So, is the file temporarily saved during the download and then deleted if the download is canceled? I can not find the file there, although when downloading other than this: safeToDelete.tmp