Resuming upload session results in 400 Upload session not found

[AnnieChmarak]

Sometimes I have 503 response while uploading a file. I use a resumable upload, so after such response I just try to resume upload session. In most cases session resumes successfully and the file upload continues. But sometimes I recieve 400 response when I try to resume the session.
I want to understade what am I doing wrong. Maybe it is a bug?

Here is a truncated and simplified Fiddler log. Let [FILE_CONTENT] be file content of the specifyed length and [UPLOAD_URL] be the url returned from POST request.

Creating upload session

request

POST https://api.onedrive.com/v1.0/drive/root:/92_32/stress/_stress_od_fiddled/1-2017%20September%2008%20(11.42)_full/.Index/20000/.i/11757.txt.i:/upload.createSession HTTP/1.1

{"item":{"@name.conflictBehavior":"replace","name":"11757.txt.i"}}

reponse

HTTP/1.1 200 OK

{"@odata.context":"https://api.onedrive.com/v1.0/$metadata#oneDrive.uploadSession","uploadUrl":"[UPLOAD_URL]","expirationDateTime":"2017-09-15T05:49:22.733Z","nextExpectedRanges":["0-"]}

Upload a file

request

PUT [UPLOAD_URL] HTTP/1.1
Content-Range: bytes 0-173/174
Content-Length: 174

[FILE_CONTENT]

response

HTTP/1.1 503 Service Unavailable

{"error":{"code":"serviceNotAvailable","message":"Service is temporarily unavailable"}}

Asking about upload state

request

GET [UPLOAD_URL] HTTP/1.1

response

HTTP/1.1 200 OK

{"expirationDateTime":"2017-09-15T05:49:22.733Z","nextExpectedRanges":[]}

Continue uploading file

request

PUT [UPLOAD_URL] HTTP/1.1
Content-Range: bytes 0-173/174
Content-Length: 174

[FILE_CONTENT]

response

HTTP/1.1 400 Bad Request

{"error":{"code":"itemNotFound","message":"Upload session not found","innererror":{"code":"uploadSessionNotFound"}}}

[ificator]

It appears as though request that was attempting to commit the final block partially succeeded - it got far enough to coalesce the blocks but not far enough to persist the result. This essentially leaves the session in an unrecoverable state. It's obviously a bug on our side, and we're tracking a fix, however there isn't much you can do other than trying the whole process again.

[patrick-rodgers]

As part of a repository clean up effort we are closing older issues. If this issue remains, please: open a new issue, reference this issue, and provide any additional details that may help in resolution. Thank you for your understanding as we work to improve our responsiveness.