Bug Description
Cannot disable "Unhealthy target connection termination" once enabled via service.beta.kubernetes.io/aws-load-balancer-target-group-attributes
Steps to Reproduce
- Set
service.beta.kubernetes.io/aws-load-balancer-target-group-attributes: target_health_state.unhealthy.connection_termination.enabled=false,target_health_state.unhealthy.draining_interval_seconds=30
- Check that the attributes have been updated in the NLB listeners
- Set
service.beta.kubernetes.io/aws-load-balancer-target-group-attributes: target_health_state.unhealthy.connection_termination.enabled=true to restore "Unhealthy target connection termination" to the default value of On
- Error!
Expected Behavior
"Unhealthy target connection termination" is restored to default value of on
Actual Behavior
Controller produces error logs
{
"level": "error",
"ts": "2026-03-27T10:44:15Z",
"msg": "Reconciler error",
"controller": "service",
"namespace": "<redacted>",
"name": "<redacted>",
"reconcileID": "fb74de70-370f-4699-a8cf-73ef7109bfd3",
"error": "operation error Elastic Load Balancing v2: ModifyTargetGroupAttributes, https response error StatusCode: 400, RequestID: 6f933ac7-86fb-4cef-87cf-744255993e14,
InvalidConfigurationRequest: You cannot have both 'target_health_state.unhealthy.connection_termination.enabled' and 'target_health_state.unhealthy.draining_interval_seconds' enabled on a target group"
}
Regression
Not sure
Current Workarounds
ClickOps
Environment
- AWS Load Balancer controller version: v3.1.0
- Kubernetes version: v1.33.8-eks-3a10415
- Using EKS (yes/no), if so version?: Yes v1.33.8-eks-3a10415
- Using Service or Ingress: Service
- AWS region: eu-west-1
Contribution Intention (Optional)
Additional Context
Bug Description
Cannot disable "Unhealthy target connection termination" once enabled via
service.beta.kubernetes.io/aws-load-balancer-target-group-attributesSteps to Reproduce
service.beta.kubernetes.io/aws-load-balancer-target-group-attributes: target_health_state.unhealthy.connection_termination.enabled=false,target_health_state.unhealthy.draining_interval_seconds=30service.beta.kubernetes.io/aws-load-balancer-target-group-attributes: target_health_state.unhealthy.connection_termination.enabled=trueto restore "Unhealthy target connection termination" to the default value of OnExpected Behavior
"Unhealthy target connection termination" is restored to default value of on
Actual Behavior
Controller produces error logs
{ "level": "error", "ts": "2026-03-27T10:44:15Z", "msg": "Reconciler error", "controller": "service", "namespace": "<redacted>", "name": "<redacted>", "reconcileID": "fb74de70-370f-4699-a8cf-73ef7109bfd3", "error": "operation error Elastic Load Balancing v2: ModifyTargetGroupAttributes, https response error StatusCode: 400, RequestID: 6f933ac7-86fb-4cef-87cf-744255993e14, InvalidConfigurationRequest: You cannot have both 'target_health_state.unhealthy.connection_termination.enabled' and 'target_health_state.unhealthy.draining_interval_seconds' enabled on a target group" }Regression
Not sure
Current Workarounds
ClickOps
Environment
Contribution Intention (Optional)
Additional Context