Comment 7 for bug 739433

Before making any changes to the database schema, I created this functional test:

    def test_size_greater_2G(self):
        """
        A test against the actual datastore backend for the registry
        to ensure that the image size property is not truncated.

        :see https://bugs.launchpad.net/glance/+bug/739433
        """

        self.cleanup()
        api_port, reg_port, conf_file = self.start_servers()

        # 1. POST /images with public image named Image1
        # attribute and a size of 5G. Use the HTTP engine with an
        # X-Image-Meta-Location attribute to make Glance forego
        # "adding" the image data.
        # Verify a 200 OK is returned
        cmd = ("curl -i -X POST "
               "-H 'Expect: ' " # Necessary otherwise sends 100 Continue
               "-H 'X-Image-Meta-Location: http://example.com/fakeimage' "
               "-H 'X-Image-Meta-Size: %d' "
               "-H 'X-Image-Meta-Name: Image1' "
               "-H 'X-Image-Meta-Is-Public: True' "
               "http://0.0.0.0:%d/images") % (FIVE_GB, api_port)

        exitcode, out, err = execute(cmd)
        self.assertEqual(0, exitcode)

        lines = out.split("\r\n")
        status_line = lines[0]

        self.assertEqual("HTTP/1.1 200 OK", status_line)

        # 2. HEAD /images
        # Verify image size is what was passed in, and not truncated
        cmd = "curl -i -X HEAD http://0.0.0.0:%d/images/1" % api_port

        exitcode, out, err = execute(cmd)

        self.assertEqual(0, exitcode)

        lines = out.split("\r\n")
        status_line = lines[0]

        self.assertEqual("HTTP/1.1 200 OK", status_line)
        self.assertTrue("X-Image-Meta-Name: Image1" in out)
        self.assertTrue("X-Image-Meta-Size: %d" % FIVE_GB in out)

I got a pass on this test without any changes to the schema. Perhaps it's due to me being on a 64-bit machine, though, which makes SQLite's Integer columns 64-bit wide?

Donal, could you add that above test case to /tests/functional/test_curl_api.py and run on your machine? Let me know the results. Thanks!