GKE ingress stuck in creating after deploying ingress 1.11.5.gke.5

[Arconapalus]

I also am running into this issue on 1.11.5-gke.5. The creating ingress is stuck and I have deleted and recreated the ingress. Should I delete node and cluster and recreate at 1.10.7.gke.2?
https://cloud.google.com/kubernetes-engine/docs/tutorials/http-balancer
I followed this guide up to the point of apply -f basic-ingress.yml and ingress is stuck creating,
I waiting for an hour then deleted and re-deployed

[rramkumar1]

Are there any events on your Ingress?

You can check by doing:

kubectl describe ing [NAME] -n [NAMESPACE]

[Arconapalus]

events are blank
Would a VPC and subnet interfere?

[rramkumar1]

Can you email me your project name, cluster name, and location so I can take a look?

[Arconapalus]

sent

[rramkumar1]

@Arconapalus Discussed offline but it appears that the L7 Load Balancing Add-On was not turned on for your cluster.

Please close this bug once you have turned on the add-on and verified everything works on your end.

[zvictor]

it appears that the L7 Load Balancing Add-On was not turned on for your cluster.

I was running into the same problem. I am part of the closed preview of knative in gcloud and the instructions provided in https://cloud.google.com/run/docs/gke/setup made me accidentally disable the load balancer.

I won't quote the precise command here because that page is tagged as Confidential Material 🙄, but just running the instructions provided there will replace the list of add-ons being added to your cluster instead of adding new ones, as some would have expected.

It is clearly stated in the documentation found at gcloud beta container clusters create --help:

 --addons=[ADDON,...]
    Default set of addons includes HttpLoadBalancing,
    HorizontalPodAutoscaling. Addons
    (https://cloud.google.com/kubernetes-engine/reference/rest/v1/projects.zones.clusters#AddonsConfig)
    are additional Kubernetes cluster components. Addons specified by this
    flag will be enabled. The others will be disabled. ADDON must be one
    of: HttpLoadBalancing, HorizontalPodAutoscaling, KubernetesDashboard,
    Istio, NetworkPolicy, CloudRun.

Even though I don't necessarily disagree with the chosen api, I can foresee many others experiencing the exact same problem and getting stuck with ingress while trying to deploy other features like knative.

I would suggest making improvement to detect the presence of the required add-on. Being in the dark with no events or errors thrown, with only this issues repository to help is not a comfortable situation.

[rramkumar1]

@zvictor Thanks for the investigation! I was wondering why we had so many issues related to the add-on not being on...

I would suggest making improvement to detect the presence of the required add-on. Being in the dark with no events or errors thrown, with only this issues repository to help is not a comfortable situation.

This is tricky because without the add-on being enabled, the controller cannot even be deployed and thus cannot emit events. However, I 100% agree that this is a bad UX. I'm going to file a separate issue in this repository for better tracking and update there once we have a solution.

In the meantime, I will also reach out to the owner of that documentation so we can get it fixed.

[rramkumar1]

@zvictor Docs are fixed now.

[caquino]

I think I have a different issue but with a similar outcome, I'm seeing my ingress stuck in create.

I'm running 1.13.6-gke.0 and I've checked addonsConfig on the cluster, and http load balancing is not set to "disabled".

I can't see any events other than "ADD" on my describe ingress.

Also when I check the load balancer on the console I see that it has no frontend attached to it, which I guess is waiting for the ingress to be provisioned.

My namespace + ingress name has 35 characters, which probably doesn't trigger the 55 chars bug

Is there anything else I can do to debug this issue?

[rramkumar1]

@caquino Can you please file a separate issue in this repo?

[manast]

@caquino did you create a separate issue? I think I am running into exactly the same issue. Stuck Ingress and empty frontend on the loadbalancer, also no ephemeral IP assigned to the ingress.

I actually tried to delete it and create it a few times, for no use:

[manast]

as an extra piece of information, even trying to create new ingress gets stuck now. Seems like something broke in the cluster and it is not possible to create new ingress, If I have to guess I think it gets stuck in the creation of the ephemeral IP.

[rramkumar1]

@manast Can you please file a separate issue?

[manast]

@rramkumar1 yes. I am doing some more tests before I fill the issue.

[manast]

I upgraded the cluster to version 1.13.6 and the issue seems to be resolved.

[salmansheikh5]

Hi everyone, I was having the same problem; but the solution I found was that I had to enable the "HTTP load balancing" Add-on in the cluster's add-ons settings.

When the HTTP Load Balancer was disabled it gave a warning that the Kubernetes ingress object may not work properly.

It took some time for the cluster of update, after that I deleted the stuck Ingress object and recreated it. And it worked. I hope my solution helps.

P.S
I also upgraded my cluster to 1.12.9-gke.7 (as stated above). But that solution didn't work for me.

[djsell]

I wasted a long time trying to figure out why my Ingress wasn't working, and @muhammadsalmansheikh fix worked. This should really be brought to attention on the Services & Ingress page here https://console.cloud.google.com/kubernetes/discovery

It'll save a lot of people time if it just displays a warning like, "Ingress isn't available because you don't have the HTTP Load Balancer add-on enabled."

Thanks

[bowei]

It's an unfortunate aspect of the spec, but many people run their own Ingress provider (e.g. ingress-nginx). It's difficult to tell if disabling and having Ingress objects is around is intentional or not from the provider standpoint.