Why it exists
The default NGINX ingress error pages are blank and unhelpful. This serves custom pages for the common ingress failure codes instead, so a misrouted request or a downed backend fails with something a person actually reads rather than a bare status code.
What I learned
Error pages are one of those details that only matter when something is already broken, which is exactly when polish tends to get skipped. Building them in advance means the worst moment for a user isn't also the ugliest one.
Status
Active, running in front of the homelab Kubernetes cluster.