You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The files basically declare what types of traffic can be accepted by the service, and where to send those network requests.
All of our service ingresses have extremely simple ingress rules: all traffic (HTTP path /) is sent to the service running on port 80 (Internet Service www).
Also, we only use a single ingress controller -- ingress-nginx -- throughout our entire cluster.
The two changes I'd suggest are:
Instead of declaring an HTTP-path based rule, use a default backend in every service ingress definition file.
Explicitly declare ingress-nginx as the ingress controller to be associated with each service ingress.
Yes, in theory we could annotate it once as the default, and rely on that working for every service ingress definition file implicitly.. but it seems better and slightly safer long-term to make the declaration explicit.
During the recent upgrade of our
ingress-nginxcontroller in #57, I noticed two deficiencies with the per-service ingress definition files.Each service we run in Kubernetes has one of these files; for example the
apiservice currently has: https://github.com/openculinary/api/blob/a488c6a32c14ce8e13e8431231dd2b1dca113989/k8s/web-ingress.yamlThe files basically declare what types of traffic can be accepted by the service, and where to send those network requests.
All of our service ingresses have extremely simple ingress rules: all traffic (
HTTP path /) is sent to the service running on port80(Internet Servicewww).Also, we only use a single ingress controller --
ingress-nginx-- throughout our entire cluster.The two changes I'd suggest are:
Instead of declaring an HTTP-path based rule, use a default backend in every service ingress definition file.
Explicitly declare
ingress-nginxas the ingress controller to be associated with each service ingress.