I'm currently doing a research on this matter. How does HTTP 404 affects the scripts and the test results? What is the best solution to capture it in our scripts? Our team uses VUgen and Performance Center.
Thanks for replying. Based on our experience, this error is returned whenever we request for a web page that has reference to other files or pages which no longer exist. As an example, a web step in the script may request for a web page through a URL. The returned page may contain several images (*.gif). Some of these may still be present in the server(s), others may not, and still others may have been renamed. We configured our scripts to continue when non-critical resource errors were encountered and have them as "warning message" and not "error message". We are concerned about the measurements that HTTP 404 errors would affect in the test result.
Since the application you are testing references files that are no longer there or renamed, then that is what will happen when real users access the system. My thoughts are that the testing is valid.
I have run into this in the past and suggest that you compile a list of all 404s you see and what pages they are coming from. Give this list to the developers and ask them to update the references in the code. They may not even realize that the files they are including are not there or have been renamed.
I agree with Terri that the test is valid from the standpoint that it is the same load a real user would generate with the existing code. HTTP 404 errors as such will not negatively impact the test results.
But if the incorrect code is going to be fixed, this will have some amount of impact on the actual page load time because the server will actually be handleing more throughput. It would probably be neglagable for a few small .gif files, but if it is lots of files and/or they are large files, then it would skew the results more.
A problem is a difference between what is perceived and what is desired, that
we want to reduce (Dewey 1933)