Google Cloud Storage upload 10gb filesize error in Cyberduck - google-cloud-storage

Question

I'm trying to upload a big file (9gb) to google storage using Cyberduck.
The login and transfer with small files work. However for this file I'm getting the following error:
GET / HTTP/1.1
Date: Wed, 30 Apr 2014 08:47:34 GMT
x-goog-project-id: 674064471933
x-goog-api-version: 2
Authorization: OAuth SECRET_KEY
Host: storage.googleapis.com:443
Connection: Keep-Alive
User-Agent: Cyberduck/4.4.4 (Mac OS X/10.9) (x86_64)
HTTP/1.1 200 OK
Content-Type: application/xml; charset=UTF-8
Content-Length: 340
Date: Wed, 30 Apr 2014 08:47:35 GMT
Expires: Wed, 30 Apr 2014 08:47:35 GMT
Cache-Control: private, max-age=0
Server: HTTP Upload Server Built on Apr 16 2014 16:50:43 (1397692243)
Alternate-Protocol: 443:quic
GET /vibetracestorage/?prefix=eventsall.csv&uploads HTTP/1.1
Date: Wed, 30 Apr 2014 08:47:35 GMT
x-goog-api-version: 2
Authorization: OAuth SECRET_KEY
Host: storage.googleapis.com:443
Connection: Keep-Alive
User-Agent: Cyberduck/4.4.4 (Mac OS X/10.9) (x86_64)
HTTP/1.1 400 Bad Request
Content-Type: application/xml; charset=UTF-8
Content-Length: 173
Date: Wed, 30 Apr 2014 08:47:36 GMT
Expires: Wed, 30 Apr 2014 08:47:36 GMT
Cache-Control: private, max-age=0
Server: HTTP Upload Server Built on Apr 16 2014 16:50:43 (1397692243)
Alternate-Protocol: 443:quic
Am I missing anything? Thanks.

Answer 1

According to that log you posted, you're placing a GET to "https://storage.googleapis.com/vibetracestorage/?prefix=eventsall.csv&uploads".
I don't know what that "uploads" parameter tacked onto the end is, but it's not a valid parameter for requesting a bucket listing (which is what that request does).
If you place that request by hand, you'll see this error:
<?xml version='1.0' encoding='UTF-8'?><Error><Code>InvalidArgument</Code><Message>Invalid argument.</Message><Details>Invalid query parameter(s): [uploads]</Details></Error>
Also, as a general point of good practice, do not post logs that contain your full Authorization header. That is a very, very bad idea. You may want to delete this question, although those credentials will expire (and perhaps already have).

Answer 2

This is an interoperability issue. In Cyberduck, when connected to S3, multipart uploads are supported as defined by Amazon S3. The request with the uploads parameter is used to find already existing in progress multipart uploads for the same target object that can be resumed.
Make sure to choose Google Storage and not S3 in the protocol dropdown list in the bookmark or connection prompt. Multipart uploads should then be disabled.