kubernetes: AWS LoadBalancer doesn't handle updates correctly

Is this a BUG REPORT or FEATURE REQUEST?:

/kind bug

What happened: After adding service.beta.kubernetes.io/aws-load-balancer-type: nlb annotation to LoadBalancer service on AWS a new network load balancer gets provisioned, but the old one is not deleted and status.loadBalancer.ingress service field is not updated with the new address.

What you expected to happen: Old load balancer is deleted. status.loadBalancer.ingress field is updated.

How to reproduce it (as minimally and precisely as possible):

Create LoadBalancer on AWS without service.beta.kubernetes.io/aws-load-balancer-type annotation;
Add service.beta.kubernetes.io/aws-load-balancer-type: nlb annotation to created service once it ensures load balancer.

Anything else we need to know?:

Environment:

Kubernetes version (use kubectl version): v1.9.6
Cloud provider or hardware configuration: AWS
OS (e.g. from /etc/os-release):
Kernel (e.g. uname -a):
Install tools: kops
Others: image kope.io/k8s-1.9-debian-jessie-amd64-hvm-ebs-2018-03-11

About this issue

Original URL
State: closed
Created 6 years ago
Comments: 22 (13 by maintainers)

Most upvoted comments

@metral I am trying to get approver to merge this PR and cherry-pick to kubernetes 1.11. Should be done soon. @micahhausler

Jeffwan on Feb 15, 2019

@bnutt It’s an unsupported behavior for now. service.beta.kubernetes.io/aws-load-balancer-type should be treated as immutable. If you need to change it, you should delete the service and recreate.

M00nF1sh on Dec 8, 2020

Hmm this is an interesting edge-case.

We don’t re-create an ELB (or NLB) right now if you switch from internal to external (see #60393 and #64251), and this feels similar. We could probably manage this, but we’ll bring it up at the next SIG-AWS on Friday

micahhausler on Jul 26, 2018