Search code examples
pythondjangohttp-streaming

How do I prevent a 500 CORS error when attempting to create and download a csv file containing a large data set using python's StreamingHttpResponse?


The function runs and works perfectly fine for a small data set, but for a large data set, the following error is produced:

(URL is hidden in the red)

enter image description here

A 500 CORS error. As you can see from the code below, I've tried adding multiple response headers to the response (StreamingHttpResponse object) with a focus on extending the amount of time the function can run/execute/query before being timed out. I've also tried adding headers to handle the cross site issue which I don't believe is a problem in the first place because the function works perfectly fine with a small dataset:

 def create_and_download_csv(self, request):
    qs = self.filter_queryset(self.get_queryset())
    serialized = self.serializer_class(qs, many=True)

    # mapping csv headers to columns
    headers = {
        "column_header": "Column Header",
        "column_header": "Column Header",
        "column_header": "Column Header",
        "column_header": "Column Header",
    }

    response = StreamingHttpResponse(
        renderers.CSVStreamingRenderer().render(
            serialized.data,
            renderer_context={'header': headers.keys(), 'labels': headers}
        ),
        content_type="text/csv",
    )
    response["Access-control-Allow-Headers"] = "*"
    response["Connection"] = "keep-alive"
    response["Content-Disposition"] = 'attachment; filename="status.csv"'
    response["Access-Control-allow-origin"] = "*"
    response["Keep-Alive"] = 200
    response["Timeout"] = 100
    print(response)
    return response

Perhaps I'm placing the headers in the wrong place? Or could it be that the docker container the project runs on is where I should configure the timeout value? Please help if you have an idea.


Solution

  • Turns out the issue was the timeout set on Gunicorn and Kong which are used in the deployment/hosting/serving of the project. The request was timing out because the workers were killed off before they gave a valid response.