Great question, since I am experiencing similar problems. I wanted to start with a link to a page of a Google developer doc (which I am sure you viewed):
As they noted there, "App Engine support for PageSpeed ​​is an experimental, innovative and fast-paced new feature for Google App Engine. Unfortunately, at the edge of bleeding, we can make backward-incompatible changes to App Engine support for PageSpeed "We will let the community know when this feature is no longer experimental."
So I'm not surprised to hear that you experienced changes in performance that were somewhat arbitrary. To paraphrase the developer’s documents, I see the following in terms of testing (since this is what I am trying to do for my own application). To work with things, you can try to be very intentional and try different test configurations (which, I believe, already exist). This, said, if you haven’t done this, and you want to try a new configuration, perhaps to test some "risky" optimizations, you can expect that this can be done in the test version of the application, while your users continue to use the default application. Version. Unfortunately, since updating the configuration of the test version applies to all versions, the "risky" settings apply to the version that your users also use. Instead, to try these settings, you can:
Disable PageSpeed ​​for your application, update your PageSpeed ​​configuration and view your site with the PageSpeed ​​chrome extension
Copy the relevant parts of the application into a separate test application that has its own identifier and application versions.
If you find any specific solutions to the problem, submit them here!
source share