Landing page image v2.0.10 breaks on production deploy

[Woozl]

It has a client side error when attempting to fetch the webpack js/css chunks. The behavior is very weird, when refreshing the page it changes which resources it's trying to fetch. It almost looks like it's remembering which resources it fetched on previous deploys and tries to pull those, but they're now different hashes so the fetch returns a 404. Even stranger, if you keep refreshing, occasionally it will request the correct combinations of resources and display the page correctly.

I've verified the container is working locally and on Sterling through a direct port-forward to the service, which leads me to believe the nginx ingress controller is doing something weird. I've tried:

• Completely redeploying the umbrella chart
• Removing the ingress annotations (24hr timeouts meant for qgraph) and upgrading

Potential sources (neither seem promising):

• nginx rewrite annotation regex missing the certain files??
• host path prefix ordering? I'm assuming the prefix paths are evaluated from first to last, which would make sense with how it's set up now, with the landing page (most broad selector) listed last.

[Woozl]

This has been fixed by rearranging the ingress config: https://github.com/helxplatform/translator-devops/pull/884