Possible issue with Ingress rules not routing? about contour HOT 11 CLOSED

from contour.

@adelbertc thanks for raising the issue. I have a feeling I know what it is.

apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: hello
  labels:
    app: nelson
spec:
  rules:
    - http:
        paths:

Doesn't specify a hostname. https://kubernetes.io/docs/api-reference/v1.8/#ingressrule-v1beta1-extensions (the host: field).

Contour expects this field to be mandatory, but I'll check the api definitions for 1.8 and confirm this.

In the meantime the workaround would be to add a host: field to your ingress object.

from contour.

Yup, this was my misunderstanding. The spec says

If the host is unspecified, the Ingress routes all traffic based on the specified IngressRuleValue.

So it should become the default vhost as in your first example.

from contour.

@davecheney Ah gotcha - where can I find what the default vhost value is to test this?

from contour.

Something like this should do it

% kubectl -n heptio-contour get svc -o yaml | grep hostname
      - hostname: a4d1766f6ce1611e7b27f023b7e83d33-1465548734.ap-southeast-2.elb.amazonaws.com

from contour.

Hm I don't seem to have that, I assume this is vendor specific since I am running in OpenStack.

Is the host value intended to be a hostname through which outside clients can reach the LoadBalancer, with the idea being hitting difference hostnames can trigger different rules, e.g. host1.foo.com/bar can trigger a different rule than host2.foo.com/bar ?

from contour.

Ah sweet, thank you!

from contour.

from contour.

from contour.

Just tried it, looks like it works - thank you for such a quick response + fix!

from contour.

from contour.