We have an Elasticsearch (7.17.5) / Kibana (7.17.5) pair running within our Kubernetes Cluster (1.21.7). When accessing the Kibana UI via the cluster's API Gateway (Broadcom API Gateway 10.1.00) all of the associated resource files come in garbled (looks like instead of UTF-8 content, appears to be Unicode).
Anyway, all of the resources are found (200 OK on their Get requests) and the Kibana and Gateway logs seem fine with the requests and their content, but the browser console shows an "Uncaught Syntax Error: Illegal Character U+001B at position 0" for all of the CSS and JavaScript files downloaded.
I can use Kubectl port-forward directly to the Kibana service, and the pages loads fine. I can also use cUrl to request the various resources and the pull down containing the standard UTF-8 JS/CSS expected.
I'm at a loss. If it was just the API Gateway, then using cUrl to access the resource through the Gateway should have the same issues. If it was just the Kibana UI, then Kubectl port-forward should fail.
Has anyone seen something like this?
Additional data point, we have a large collection of web applications (HTML/CSS/JavaScript collections that are retrieved via the API Gateway that thus far have not been adulterated into strange Unicode sequences.
Putting the answer out there for posterity, the UI request generated in the browser had a header for "Accept-Encoding: gzip,deflate,br" and the Broadcom API Gateway doesn't cope with the br.
Resolved by having all UI requests through the API gateway strip the "br" from that header and css/js resources once again loaded as expected.