Aws-load-balancer-controller: Multiple alb-ingress-controller support

[jchoi926]

Hello,
I currently have applications across two namespaces. I've spun up an alb-ingress-controller with the --watch-namespace parameter pointing to the first namespace, which works as expected.
When I spin up the second alb-ingress-controller, the controller pod gets to the "running" state however, the alb itself does not get provisioned. The logs show the controller waiting for a leader lease (not sure what that is exactly).

`AWS ALB Ingress controller
Release: v1.1.2
Build: git-cc1c5971

Repository: https://github.com/kubernetes-sigs/aws-alb-ingress-controller.git

W0726 23:36:01.489292 1 client_config.go:549] Neither --kubeconfig nor --master was specified. Using the inClusterConfig. This might not work.
I0726 23:36:01.552747 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"={"Type":{"metadata":{"creationTimestamp":null}}}
I0726 23:36:01.553445 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"={"Type":{"metadata":{"creationTimestamp":null},"spec":{},"status":{"loadBalancer":{}}}}
I0726 23:36:01.553525 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"=
I0726 23:36:01.553769 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"={"Type":{"metadata":{"creationTimestamp":null},"spec":{},"status":{"loadBalancer":{}}}}
I0726 23:36:01.553806 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"=
I0726 23:36:01.553955 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"={"Type":{"metadata":{"creationTimestamp":null}}}
I0726 23:36:01.554383 1 :0] kubebuilder/controller "level"=0 "msg"="Starting EventSource" "controller"="alb-ingress-controller" "source"={"Type":{"metadata":{"creationTimestamp":null},"spec":{},"status":{"daemonEndpoints":{"kubeletEndpoint":{"Port":0}},"nodeInfo":{"machineID":"","systemUUID":"","bootID":"","kernelVersion":"","osImage":"","containerRuntimeVersion":"","kubeletVersion":"","kubeProxyVersion":"","operatingSystem":"","architecture":""}}}}
I0726 23:36:01.554568 1 leaderelection.go:205] attempting to acquire leader lease kube-system/ingress-controller-leader-alb...`

How do I get the second alb-ingress-controller to cycle through the provisioning of the second alb?

[Seitk]

Hi, do you end up being able to solve this issue?

[jchoi926]

At the time of our implementation, AWS had some limitations in regards to the alb-ingress-controller.
I ended up spinning up a single instance of the controller without the --watch-namespace switch. Each namespace then has its own ingress in which you specify the ingress-class. This is in turn provisions an alb with the ingress rules specified. Hope that helps.

[muthusk]

For anyone who stumbles upon this, the trick is to create a new config map ( kubectl create configmap second-ingress-controller-leader-alb -n kube-system) and specify the same in the alb-ingress-controller.yaml file with “- --election-id=second-ingress-controller-leader-alb” https://github.com/kubernetes-sigs/aws-alb-ingress-controller/blob/master/cmd/options.go also has boolean “election” and I am not sure what would be the effect of setting “election” to false.

In my setup , I have one ALB Ingress Controller on private subnet and one ALB Ingress Controller on public subnet.