Upload failed on AWS S3 with "File was not opened in write mode"

[GitRon]

Hi guys,

I'm running python 3.6.x with django 1.11.11 on gunicorn 19.7.1.

When I upload a file, I get the following error:

"File was not opened in write mode."

Sentry points me to this line:

    self.file.open(mode='ab')  # mode = append+binary
    # We can use .read() safely because chunk is already in memory
    self.file.write(chunk.read())

I already checked this issue but it was related to an old version of django.

Any ideas?

Thanks
Ron

PS: Maybe this is related to my storage of media files on AWS S3? Chunked parts are created but have size zero.

[sighmon]

@GitRon Did you find a solution?

[GitRon]

@sighmon I branched out and this is my solution: https://github.com/GitRon/django-chunked-upload

[birkanozer]

nice fix

[GitRon]

@birkanozer Unfortunately we are stuck here...

[awwester]

Unfortunately we are stuck here...

@GitRon Does that mean you weren't able to get a full working solution? I've seen some other errors downstream as well, so curious if you ended up getting it fully working.

[GitRon]

I wanted to add the general fix for using a temporary storage. But the guy who took over the project started changing way more things. Variable names etc. IMHO this is too big for now and will cause loads of problems for people using the package. We somehow got lost along the way and since some time I haven't heard or (I have to admit) said anything on this topic.

My dream scenario would be that we start where the old maintainer left of, merge in my fix, maybe I have to update the docs a bit because the fix is just also a fix for the cloud. But it can be used in other circumstances as well.

When this is one the road, we could discuss in an issue if renaming variables etc is a proper way to go.