Failing to create an Ingress shouldn't stop the controller from creating the others

[jody-frankowski]

(Not sure if that is the right place to fill this issue 😃)
It looks like the controller doesn't try to create other Ingress object when it fails for one.

See this issue for the reproduction steps.
In a nutshell: Ingresses with non-NodePort services break proper Ingresses with NodePort services.

Also (not specified in the other issue) here are the ingresses that break ingress-gce (note the ClusterIP type):

kubectl get svc
NAME                            TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)     AGE
heapster                        ClusterIP   10.15.241.123   <none>        8082/TCP    4d
[...]

Two things bug me:

• ingress-gce seems to break all together when it fails for just one Ingress
• ingress-gce seems to report errors related to other ingresses for a given one

[MrHohn]

@jody-frankowski What was your GLBC version (or k8s version)? I was unable to reproduce the issue with GLBC v1.1.0 so this might already be fixed during recent refactoring. I will include this case in our e2e test for sure.

[jody-frankowski]

The k8s version of the cluster is 1.9.6-gke.0.
I'll try to reproduce the issue with 1.10 when it's GA.

[nicksardo]

Fixed in later versions of ingress
/close