[Bug]: can't get lbtargetgroups in v1.17.0 due to v1beta1/v1beta2 issue

[charlesmelby]

Is there an existing issue for this?

• I have searched the existing issues

Affected Resource(s)

• elbv2.aws.upbound.io/v1beta1 - LBTargetGroup
• elbv2.aws.upbound.io/v1beta2 - LBTargetGroup

Resource MRs required to reproduce the bug

apiVersion: elbv2.aws.upbound.io/v1beta1
kind: LBTargetGroup
metadata:
name: proxy-ingress
spec:
providerConfigRef:
name: aws
forProvider:
name: proxy-ingress
region: ap-northeast-1
port: 443
protocol: HTTPS
targetType: instance
vpcId: ************

Steps to Reproduce

kubectl get lbtargetgroups

What happened?

It can't fetch the targets groups (or individual records) due to conversion webhook failure.

Relevant Error Output Snippet

Error from server: conversion webhook for elbv2.aws.upbound.io/v1beta1, Kind=LBTargetGroup failed: cannot convert from the spoke version "v1beta1" to the hub version "v1beta2": cannot apply the PavedConversion for the "aws_lb_target_group" object: failed to convert the source map in mode "toEmbeddedObject" with source API version "*", target API version "v1beta2": value at the field path stickiness must be []any, not "map[string]interface {}"

Crossplane Version

1.18.0

Provider Version

1.17.0

Kubernetes Version

v1.28.9+rke2r1

Kubernetes Distribution

rke2

Additional Info

The AWS target group IS successfully created. However,

• I can't get the target group information or status
• targetGroupArnRef in LBTargetGroupAttachment never gets filled in, so I had to manually copy targetGroupArn from AWS instead

[turkenf]

Hi @charlesmelby,

I could not reproduce the issue. Do you encounter this issue constantly or sporadically? Could you please restart the relevant pod and check again?

> k get lbtargetgroup.v1beta1.elbv2.aws.upbound.io/test
NAME   SYNCED   READY   EXTERNAL-NAME                                                                                    AGE
test   True     True    arn:aws:elasticloadbalancing:us-west-1:153891904029:targetgroup/proxy-ingress/6047e967428e9f70   64s
> k get lbtargetgroup.v1beta2.elbv2.aws.upbound.io/test
NAME   SYNCED   READY   EXTERNAL-NAME                                                                                    AGE
test   True     True    arn:aws:elasticloadbalancing:us-west-1:153891904029:targetgroup/proxy-ingress/6047e967428e9f70   70s

[wotolom]

I just had a similar bug/behaviour:

The LBTargetGroup was stuck.

• Reapplying the claim (function) did not work

composed resource "tg-test123": conversion webhook for elbv2.aws.upbound.io/v1beta2, Kind=LBTargetGroup failed: cannot convert from the hub version "v1beta2" to the spoke version "v1beta1": cannot convert the map[string]any representation of the conversion target back to the object itself: cannot restore slice from map

• deleting the claim (function or directly) did not work

Error from server: conversion webhook for elbv2.aws.upbound.io/v1beta2, Kind=LBTargetGroup failed: cannot convert from the hub version "v1beta2" to the spoke version "v1beta1": cannot convert the map[string]any representation of the conversion target back to the object itself: cannot restore slice from map

This was confusing because the same claim worked fine on other clusters.

Then I deleted the deployment of provider-aws-elbv2. After the pod was back and running the error gone....

---

K8s: v1.30
Crossplane: v1.17.1-up.1
Provider (family/elbv2): v1.17.0